Ali Partovi, Tim Fist, China's Semiconductor Evolution, NVIDIA's China Business

You're watching TVPN. It is Wednesday, April 16, 2025. We are live from the temple of technology, the fortress of finance, the capital of capital, Jordan. You want to talk about what you're wearing?

I got a new jacket on.

New jacket merch.

These are the moments you never forget, right?

Yes.

First time you put on a jacket that you really love. I have, you know, I put this on this morning. This is a early edition of a jacket that will be being sent out to the TVPN network very shortly. But it turned out I posted a picture of it a while back.

Stand up a little bit. Stand up a little bit. Yeah. Okay, so we got some bezel, public 8 sleep ad quick wander. Ramp on the front, ramp on the back, ramp on the left, ramp on the right. It's ramp all you can see. And we thank you to Ramp for sponsoring the show and make it, making it possible. But speaking of ads, I wanted to run an ad for Ad Quick.

There we go.

It's kind of the jacket of the real world.

That's right.

It's the way to get your logo, express your brand out into the.

In the built environment.

So if you're looking for out of home advertising made easy and measurable, head over to adquick.com. say goodbye to the headaches of out of home advertising. Only Ad Quick combines technology, out of home expertise and data to enable efficient, seamless ad buying across the globe. Anyway, let's move on to the big news in tech. Today is clearly Nvidia with the H20. It has been restricted in China. And there's a bit of a debate emerging between stratecheri and semianalysis. Ben Thompson and Dylan Patel are. They have slightly different takes. I think they agree on a lot of what's going on with this project and with the story as it's developing. So we'll take you through some of the. The debate that's raging about how important chip bands are, what this means for Nvidia, what this means for China. And then also I wanted to do a little bit of a deep dive again on the history of semiconductors in China and how we got to this particular configuration where America is kind of dominant. China's kind of catching up. But, you know, China's been working on this for a long time. It's not like semiconductors are new and they never thought of them. They've been working on this since the 70s and they've had a variety of government programs and private enterprises, programs that have driven them to kind of where they are Today. And so Huawei goes off.

Yeah, the conversation with Aaron yesterday, which if you haven't listened to, go back and listen to Aaron Gin's segment, one of, one of my favorite guest appearances. Yeah, it was great to date on the show. And he basically said what we don't want is Huawei going around the world giving people worse but cheaper chips and, and just sort of like flooding the market and becoming dependent on Chinese chips.

Yep. And so we have some posts from the timeline to kick it off the Cobesi letter. Kobese letter. I don't know how to pronounce that. Breaking. Nvidia says the US government has banned them from selling H20 chips to China for the indefinite future. Nvidia says this will come with a $5.5 billion charge to Q1 earnings and the stock traded down 5%. In this screenshot, you can certainly pull up Nvidia stock on public.com and tell us what it's doing today. But Luke Metro quote tweets it and says probably the right decision. But Lowell at Nvidia designing a chip to get around export control laws, then being so successful that they change the laws. And so this is the nature of these chip bands. They're always evolving. And so we'll take you through some of that. But I thought it would be interesting to start with some, some history. So China's semiconductor ambitions really began in 1978. They started working on semiconductor R and D. This is during the Den Xoping leadership that we talked about yesterday. China launched economic reforms aimed at modernizing technology and industrial output, including early research into integrated circuits, which of course the US was doing. But during the Cold War, the US export control severely limited high tech transfers. And so Chinese labs were working with outdated technology. There were small state backed institutes that emerged and the first big one that everyone kind of focuses on is this 8, 63 program which is interesting. It's named after when it was launched, which was March of 1986. And so in Chinese it's 86 is the year and 3 is the month. So 863. And this prioritized strategic areas like semiconductors, telecommunications and biotech. And so the program set conditions whereby foreign firms entering China had to share to technology. Although core ip usually main garden. But this is like it goes back to the 80s, like this strategy of like oh you're coming here. Like let's, let's share some ip, share ip.

What's, what's the big deal?

Open the kimono. Yeah, right.

So universe, the design file. We won't we won't do anything with it. Yeah, you can trust, yeah, you can trust us.

Just let us hold it. Just let us hold it.

You can trust us with your ip.

Yeah, just let us hold it. Just let it, just let me hold it.

Just let me hold it for a second.

Just let me hold it for a second. Universities and research institutions expanded semiconductor curricula, training a generation of engineers who later contributed to domestic breakthroughs. And so the Chinese government in 1988 designated semiconductors a pillar industry. So this is not like some new oh, China's waking up to AI and now they're getting, this has been going on for a very long time. And they initiated early state sponsored projects and forged a bunch of joint ventures with foreign partners. And so reflecting on a nascent domestic industry, state plans began to include establishing manufacturing hubs, incentivizing technology transfers as we mentioned. So imported equipment helps set up primitive fabs operating at suboptimal process nodes. And this, this technology gap wound up persisting for decades and it exists today. The 3 nanometer node works in Taiwan, doesn't really work anywhere else.

Although Aaron was saying is like the gap is, is really real but not as significant as you know, TSMC or Nvidia would like maybe the market to believe.

Yep, yep. And so after the 8, 6, 3 project that started in March of 1986, they launched the 908 project which starts in 1990. And this was their first major attempt at creating a world class semiconductor manufacturer through state led initiatives. And so they had the state owned factory and in Wuxi and they converted it into a competitive integrated device manufacturer or what's called an idm. And so there were some partnerships with American companies like AT&T's Bell Labs. And this was supposed to speed up technology transfer despite significant limitations on core process knowledge. So they didn't have everything that they could do internally, but they would pay Bell Labs to bring over some of the tech. The project's execution was hampered by bureaucratic delays because the state owned and stated state run enterprise and there were slow, there was slow adoption of the advanced methods. And so by the time production began, the fab was already obsolete, which is like an ongoing theme now it's a little bit less relevant because if you can't, if you're trying to build a 3 nanometer node and you wind up with a 7 nanometer node like you can put those chips in cars, you can put those chips in like yes, you're not going to be on the bleeding edge of AI or phones, but you're going to be, there's a use for them, there's a use for them. You're going to put them in smart fridges and smart toasters and smart thermostats and all sorts of stuff. There's chips everywhere.

Smart home will be dumb as rocks.

And so then there's Project 909 which was designed as rapid push to build an 8 inch wafer fab that could meet international standards. Though later they ran into some trouble here. This was ordered by president Zhang Jimin, President. Project 909 set up a modern 8 inch fab in Shanghai within two years marking China's first serious stride in modern wafer production. This initially showed promise because they achieved this point 0.5 to 0.35 micron process in 1997 but soon encountered profitability issues exacerbated by the 2000s.com crash. And so this is the, this is the nature of building a fab. If you're not on the leading edge you don't get the crazy margins and so you, you have to invest a ton in Capex and then recoup that over a number of years. And so if you're not on the leading edge and you can't have those high margins that then yeah, it's basically.

Already commodity or you know, do you actually have IP?

Exactly. And so in the late 90s China focuses on the for the first time on chip design. This is what Nvidia does but also Google with the TPU and Meta Meta has a new chip as well. Amazon and pretty much everyone at this.

Point, Apple, Microsoft, but that's more quantum focused.

Yeah, I think Microsoft is probably developing like an AI accelerator chip just like everyone else but we haven't heard that much about it. All these companies are still huge buyers of Nvidia but the, the chip design is obviously different than Fab. And so this was the China's first step towards self reliance and semiconductor design. Previously they'd been licensing designs and then say hey we'll just be doing the manufacturing. But having that integration seemed critical at the time. So Huawei's Hisilicon was formed to supply Asics for telecom while companies like Godson or Longson emerged to design CPUs. And so you know, if you have a telecom tower it needs to do certain calculations. It's not entirely a general purpose cpu. It doesn't need to be able to do any program but it needs to be able to do very specific things again and again, again. And that's what an ASIC is Great for so it's essentially like an algorithm on a chip. It's how I think about ASICs. Like Bitcoin is the classic one.

Yeah. So they launched JVs at this time with, with basically foreign firms. So alongside state projects a growing number of private enterprises began designing, designing chips tailored for the domestic market. So remember this point, this is the 90s, we still had this feeling that, that China wasn't necessarily gonna destined to be an adversary. If we could turn them to the, the fantastic, incredible system of capitalism, democratic capitalism, we could you know, get along and everybody would have a great time. And that, that was like basically like top down sort of like administrators view at the time.

And so this is all pre world World Trade Organization but they joined of course in the late 90s and this spurred greater open openness and then a lot of multinational corporations love to see those. But they were also built out of like these high tech parks. So Shanghai has a high tech park that has semiconductor related research and light manufacturing going on. And this is really the key to what makes China amazing at like the iPhone City. IPhone City is not just Foxconn. It's not just the final assembly. It's that classmate manufacturer, the lens manufacturer, the chip manufacturer, camera. Yeah, all these different pieces start forming and China is learning that having a somewhat free market, somewhat decentralized process for building these industrial parks is actually the dominant philosophy. And so they let different companies compete in these segments but they wind up creating this, this like rainforest of value creation. And so the company, China's still facing challenges at this time. They never really master the high end process technologies. And most advanced production remains abro. But in 2000 this is the founding of SMIC, the semiconductor manufacturing International Corporation. Great name. And, and you've been hearing a lot about smic.

Now these names are coming back too, right? We've seen this Advanced Manufacturing Company of America.

Yeah, we saw Allen Control Systems kind.

Of fits in earlier this week. General Mac was kind of ripping SF Compute.

Oh yes.

Doing the general intelligence of New York. Of New York.

The browser company of New York. Yeah.

So this meta is dead now. Yeah, go back short.com but, but it's.

Portmanteaus actually I want to go back to Portmanteaus. Bitly. These like cute little feas ly. We got to bring those back now.

Tbpnly.

Yeah. So SMIC was set up as a pure play foundry and it was modeled directly on Taiwan's tsmc. So they saw what, what Morris Chang was doing in Taiwan with support of the government. Remember the government came in back TSMC and, and China's just like, yeah, let's just do that with SMIC. And they've been doing that for 25 years now. And they're, and they're like slowly going down the learning curve, increasing the process nodes. And so they progress from 180 nanometer to 90 nanometer nodes within a few years. Remember we're down at 3 nanometers now.

Yeah. And the interesting thing, you have these, you know, basically like state run corporations that have bureaucracy. They aren't hyper efficient but, but they are strategically important enough to the country that they just get totally, they just like become these sort of like snowballs of capital and over time sort of brute force their way into some amount of relevancy.

Yeah. So SMIC is moving down the process nodes. They go from 180 to 90. Remember now we're down at like 3 nanometers. And some of these terms are more like marketing terms now like 4K TVs have different, different specs and whatnot. At a certain point I believe the, the width of a silicon atom is one nanometer. And so you can't go, you can't pack them closer. You could never have like a 0 nanometer chip because it wouldn't work. So that's where like quantum comes in.

Never, Never.

Yeah, I don't know, I'm not a physicist. Like it's totally possible.

Ben's working on something there.

Yeah, we'll see. Someone's cooking something up. Someone's going to raise a mango seed for that idea. Yeah, 1nm chips. 0nm chips.

We're only one mango seed away from it.

Yes, yes.

Never lose faith.

But, but at this point like SMIC is starting to position itself as a, found as a serious foundry, not at the leading edge, but at the trailing edge. And so in the early 2000s there's a boom in Chinese chip design firms which lays this foundation for more domestic semiconductor innovations. And companies are embracing the fabless model, like what happened in America with Nvidia. So Spread Trim and Godson started designing chips specifically for mobile and computing applications. And, and just in general, China is improving domestic R and D, state sponsored initiatives, university programs, that type of stuff. And so sure, corporate espionage as well. And so in 2004, policymakers acknowledged that domestic production was meeting only a fraction, around 10% of China's enormous semiconductor demand. So they're making a lot of electronics, obviously they want to be a major player there, but they keep having to buy 90% of the semiconductors Abroad and that is seen as, as unacceptable. And so they want to continue to compete in semiconductors. So China's domestic chip production capacity was far from adequate to service its $40 billion annual chip consumption. So by comparison, China's only producing $4 billion in annual chips, but they're consuming 40 billion.

And so trade imbalance.

Yeah, yeah. I mean they did have a trade imbalance in chips and they wanted to take that very seriously. And that's how, kind of how we got here. And so they, and so they did some policy reforms, they set up some frameworks for future government supported R and D. They're just continuing to invest government dollars in new semiconductor projects. And so from 2005 to 2007, you're seeing a maturation of the global supply chain, foreign investment and technology transfers. Intel actually expands into China and this is all accelerating these modern semiconductor practices, practices. And China is going down the learning curve. So intel built key assembly and testing facilities, including a major plant in Chengdu and later a 300 millimeter fab in Dalian announced in 2007, marking significantly significant technology transfers to Chinese engineers. The spread of multinational semiconductor companies operations in China bolstered local expertise and catalyzed market growth. Although Nvidia was founded in 1993 and initially focused on the US gaming market, by the mid 2000s it began establishing R and D centers in Shanghai to support increasing demand in China.

Yes, this is the start of a long, complicated relationship.

Yeah, I mean these are global companies and they have employees in China. And so turning off the faucet, it sounds really easy when it hits the front page of the Wall Street Journal. But it is much more complicated than that because these, all these entities are super intertwined. So then of course the global financial crisis hits. 2008, 2009, Chinese government is dropping huge stimulus packages and funding more and more semiconductor expansion. They don't want their fabs to go out of business if there's a hiccup in demand for the final. And so they are just continuing to back these semiconductor projects because they know that it's critical to their industry. So effectively like a bailout for the Chinese semiconductor industry during the global financial crisis.

Honestly, amazing, you know, foresight all around.

Yeah, yeah, you shouldn't let that die. It's a very big mistake. And I mean people have argued that America has made mistakes during downturns. Not supporting certain industries, letting them de, industrialize or move abroad. So in around 2014, the Chinese government established the National Integrated Circuit Industry Investment Fund, or as they call it the big fund $20 billion going into semiconductor sector all about innovation that is led in China. Indigenous innovation, as they call it. The BIG Fund provided capital for upgrading fabs, design houses and even research into next generation semiconductors. And many of China's now leading semiconductor companies like SMIC and Huawei's high Silicon received funding boosts to accelerate their competitiveness. Around the same period, Nvidia set up R and D centers in China and deepen partnerships with companies like Baidu and Alibaba. 2015, we, we talked about this with Xi Jinping. The Made in China 2025 plan was unveiled and we didn't dig into the details of that, but within the Made in China 2025 plan they, they specifically called out semiconductors, said we need self sufficiency in semiconductors. And this was also while the US started restricting chip imports. And this affected intel and Qualcomm most directly. And so the specific line in made in China 2025 is that they wanted to hit 70% of domestic production of chips by 2025. And I'm not exactly sure where they are. Maybe we'll get to that later. But you can see how serious they're taking this. At one point they're at 10% and they're like, okay, but we have a firm plan and we're willing to invest government dollars to get to 70%. This is very important. And so how, and.

It'S an interesting dynamic, right, because you can imagine, you know, these are, these are state mandates and they're goals, but there's real people's livelihoods on the line to deliver those. So you can imagine the pressure to actually deliver, deliver against these timelines is intense. This is not like, you know, a us, you know, corporation saying, you know, we're hoping to reduce, you know, energy, you know, use more clean energy in 2035. It's like we are going to do this one way or another. And they would cycle through executives in order to achieve these goals.

Right.

And so even though it's bureaucratic, there's still, I think this intense pressure that in many ways is greater than the pressure that even individual companies feel from their VCs.

Right.

If you have the party saying like this is strategically important to the country and you need to achieve this and this is the goal we're setting as a country, then you better figure out a way to do it. Right.

Yep. Yeah. And so Huawei is really emerging as a champion around this time. They actually have an internal chip design arm, Hisilicon. You can think about this like the relationship between Apple and the Apple Silicon team. So we think of Huawei as making phones, they make a lot of stuff now, but they also design chips specifically. And the Huawei AI chip is the one that's the focus of Dylan Patel's analysis and semi analysis yesterday or today. So this is also starting to become the beginning of what becomes chip bans, chip acts, restrictions. The U.S. department of Commerce block transactions with entities such as ZTE. And this was the beginning of this confrontational phase in semiconductor export controls. And so in 2016, 16 Chinese companies, they're getting a lot of state funding and there's clearly a lot of strategic urgency. They're starting to attempt cross border mergers and acquisitions. And the US regulatory system is saying hold on, let's not do that. So Tsinghua Uni Group acquired domestic players, Spread Trim, RDA Microelectronics we talked about before in a drive to build integrated chip manufacturing and design ecosystems. And this is all waking up the US to the increasing need to restrict technology transfer and stop Chinese firms from buying. Little late but, but this is. Who was saying this? Like, you know, America's like the sleepy student that wakes up and then just goes for it.

Yeah, yeah, yeah. You wake up and you're sitting on the couch, you put on 50 extra pounds and you're like wow, I gotta get in shape.

Yeah.

And then we do it. Historically we've done it, we've done it.

And so that's what's happen. Woke up on the couch, a little chaotic.

We sprinting into the battle for you know, a couple decades.

Couple decades.

And now we're going to get in the gym.

Yeah, now we're going to the gym. So in 2017, China unveiled the New Generation AI Development Plan and the global AI revolution took off. US policymakers intensified reaction restrictions that would limit access to advanced chip technologies. So a lot of this was driven by Google's AlphaGo success when they beat Lisa Doll and everyone in tech and venture was starting to see the power of artificial intelligence. And so AI investment was increasing dramatically. This wasn't, I mean obviously OpenAI was working on like super intelligence and talking about AGI, but even X that people were thinking about, well AI is just an incredibly valuable technology even in the context of like for you feeds and automated data analysis and just back office like automation, all that stuff is like AI is going to be an increasingly important part of the economy aside from all the AGI ASI conversation. So the US administration blocks several Chinese tech deals, including targeting companies on the entity list. And, and this is kind of like the precursor to the controls on AI chips that we've seen more recently.

And it's so wild to put this in the context of OpenAI actually starting in 2015 and being a hyped but still under the radar organization because it was, it wasn't in the headlines constantly, right?

Totally.

It was really a research organization.

And that was the branding.

Yeah, that was a branding, really cool initial branding. But you can guarantee that China was paying attention to this, right? It's like, okay, if Elon and Sam Altman and all these people are involved in this project, there must be something significant.

I remember there's this photo that Elon shares of Jensen Huang delivering one of the first kind of AI specific Nvidia servers to the OpenAI team in like 2017, something like that. And I remember seeing posts on then Twitter about from like Greg Brockman being like, oh, we just got the latest box from Nvidia. Thank you so much. To them this thing costs like a couple hundred grand. But it was really important because it, it had enough, I believe it had enough memory on the single server chip that you could put basically the entire text of Reddit in it in memory and then train on that. And that's where like the GPT1, GPT2s were trained. They weren't even trained on like massive clusters.

Pretty sure they were trained on 2017. I'm assuming that Greg and Sam already had a sense that they would go need millions of GPUs.

So I think they're like getting scale pilled at this time and kind of discovering the scaling laws but not maybe fully realizing them. But globally I think people are just realizing that the increasingly Nvidia is going to run away with the game when it comes to AI development. Like you can look at that server and, and, and there was never a question that a team like OpenAI would train on anything but Nvidia. And then that just had this compounding advantage that when they want to scale up and they want to go from one rack, one rack of chips to 10 racks to an entire data center, it's just going to be Nvidia all the way because they're writing code that runs on cuda. Cuda is this like ecosystem that Nvidia's built and it's going to be very hard for them to port to something else. Even if another company is better on like a flop per dollar ratio, Nvidia is just kind of running away with it. And this is, this is something that it's a meme that goes global. So China's major cloud provider start ramping up orders for Nvidia chips all the way back in 2017. And so in 2018, this is kind of the beginning of that first trade war that we talked about, the one that's kind of like a nothing burger now, but was kind of dramatic at the time. And so there were tariffs and export controls that we talked about. But this severely impacted companies like zte and it forced Nvidia to recalibrate its product strategy. In April of 2018, the US imposed tariffs on hundreds of billions of dollars worth of Chinese goods, and China retaliated with similar measures. The harsh punishment of ZTE forced companies to re examine resilience on US Components. And Nvidia, the gaming and AI division, saw robust demand, solidifying the company's importance in both sectors. But obviously there's mounting regulatory pressure.

And so we should go back and find interviews with Jensen around that time. He's always just been caught in this tough spot, right, Because Nvidia had offices in Shanghai, it had R and D in Shanghai.

All of that doesn't matter. It just has shareholders that want to maximize value. And so like, yeah, like their money's green. Yeah, as long as they convert it to dollars, I guess. But you know, like you want to expand your market and, and every tech company tried to go to China. Many of them got blocked. Nvidia was just the one hyperscaler that really didn't get blocked because they were just selling in the hardware. And China wasn't worried about that, but they were worried about Google, they were worried about Facebook and, and many other companies didn't have a good, a good opportunity, mostly because of like free speech stuff. But Nvidia, that doesn't matter because just hardware and you can run whatever you want on top of it.

So it was, it was a lowly $80 billion company in 2018 too. Was hardly even on the map.

Exactly. And so in 2019, Huawei gets added to the US entity list. This is kind of the black list. China accelerates investments in domestic chip design while Nvidia deepens its relationship with cloud giants like Alibaba and Tencent. And so being blacklisted meant that Huawei lost access to US technology. So they couldn't buy because if you're on the entity list, you can't sell or you can't do any business with any American company. And so that means that it also applies to companies like ASML that use patents that are American. And so if you can't buy a lithography machine from the Netherlands. You can't fab the chips. And so it's just a major setback when you get added to the entity list. And we've seen the opposite happen where American entrepreneurs are now being added to the Chinese version of the entity list and they can't buy products from China or they can't sell products to China. Not that they want to.

Yeah. Just give you a sense of Huawei scale. They did over $118 billion in revenue in 2024.

Wow.

As a private company.

Oh, private. Interesting.

Private. Interesting.

Yeah. I mean, they probably don't want to deal with the oversight of the is.

Simply don't want to deal with.

But, but all these restrictions basically force the high silicon unit within Huawei to be like ultra self reliant because they're just like, we can never count, like, we're probably never getting off the entity list. Like, it is pretty bipartisan. Even though it's like a Trump thing. It's like they're not expecting it to just flip. It's not like Americans are yelling like, Trump is so crazy because he put Huawei on the entity list. Like, we got to get Huawei off the entity list. Like, it's not a culture war issue. And so it's just truly not. It's like there's no sign that this would ever flip. And so as Huawei's orders dried up, the other Chinese tech giants stepped in to purchase Nvidia's GPUs. And so the, the U.S. like Nvidia maintains significant share of the Chinese data center market because the Chinese data centers, they obviously, you know, they kind of want to support Huawei, but if Huawei is way on the back foot because they don't have amazing machines to build the stuff, well, the, you know, Alibaba and Baidu, they just need to run servers, they need to do accelerated computing, AI inference, all sorts of different, you know, recommendation systems. And they're like, yeah, like we just need, we just need the best.

Yeah.

And so there was an entity list expansion that, you know, covered even more advanced chip technology. And this has just been growing and growing and growing. So in 2020, there's even more U.S. restrictions. The for the foreign direct product rule, or FDPR, was expanded to halt Huawei's access to advanced chips, whilst China simultaneously launched its second generation Big funds, little size. Going for the big fund, Nvidia attempted a $40 billion acquisition of ARM. And it was announced via this, like amid like all the turbulence in the FDPR expansion and there was more sanctions on SMIC in December of 2020. That again set back China's progression down the learning curve into these advanced nodes and made it even more important that they focus on domestic semiconductor manufacturing capabilities. So now we get to the CHIPS Act. This is like where the story usually starts, but we went a little bit further back. But the U.S. chips and Science act passed in 2021. This is a Biden administration law and this had coordinated export controls by key allied nations and it accelerated the decoupling of the US and Chinese semiconductor supply chains. And of course this was a multilateral effort.

This is almost three years after the US had organized for the arrest of the CFO of Huawei, which was the founder's daughter.

Yes, it wasn't that in like Canada, like really aggressive.

So this feels like a really long time ago at this point.

Totally.

But it's important to understand like just, just how serious kind of everybody in Washington was taking this at this point.

Yeah, yeah, yeah. And so this is specifically restricting critical chip making equipment such as ASML's, EUV lithography machines, that's extreme ultraviolet lithography machines, the most advanced machines that, that can, you know, place the, etch the silicon atoms like as small as possible. Right. And so the US is coordinating with the Netherlands and Japan because Japan also makes a lot of like lens technology and a lot of different things deeper in the semiconductor supply chain. And so all of that is now restricted. And so China in response is doubling down on, on self reliance through the 14th Five Year Plan and they have extensive subsidies for semiconductor development. Just one more five year plan, bro. Just one more five year plan, please.

No, it's actually if you're going to be, you know, if you're going to run do state run industries, you know, working in five year increments, like probably is pretty effective if they're on their, their 5th or their 14th five year plan. The five year plan model I would imagine works.

I think it's working pretty well. They like it. And so Nvidia struggled with delays in its ARM deal or arm. They were trying to buy ARM and, and they keep seeing resilience from the Chinese cloud providers even as political risks mounted. And, and so the AI hardware race is just heating up even further. And so in 2022, October 7, 2022, US export controls targeted high end AI chips. This is banning Nvidia's flagship products, the A100 and the H100 to China while forcing the development of workaround products and so the new rules block shipments of advanced chips and manufacturing equipment for nodes below 14nm for logic and 18nm for DRAM, which is the memory.

Yeah. And so what did Nvidia do? They introduced the A800 GPU, an adjusted version of the A100 with lower interconnect bandwidth to skirt the US limits without fully matching performance. So again, this is like one of those situations where people in the US I think would be like much more frustrated with Nvidia's behavior if they didn't were not Nvidia shareholders.

Yeah. This always struck me as interesting because when you think about the US government regulating interconnect bandwidth on an AI chip, that's like extremely complicated. And this is like the, when I think of the government and technology, I think about like, how does Facebook make money? Like, the Internet is a series of tubes. Like the lawmakers typically get this stuff like wildly wrong. Clearly they had some great consultants on this project because they actually got to a level where they were able to understand some of the trade offs in chip design to design a policy that targeted specifically AI training around interconnect bandwidth. Like they did hit that, but at the same time they weren't able to go deep enough to understand that deepsea could like figure out a way around the Internet, the interconnect bandwidth limitation, and still train a GPT4 class model. Right. So they were like 99% there, but not 100% there. And like way better than any layman. Especially at the time, like back in 2022, not many people could just be like, hey, you know, oh you, you know a little bit about semiconductors, like what's important right now? And people would be like, interconnect bandwidth, like, yeah, right.

But somehow 800 and Chinese firms just go, oh sweet. They just start stockpiling these things, right? They just start saying like, we'll basically take as many as you can get us.

Yep. And so in 2022, in December, SMIC gets added to the entity list. Kind of crazy, they weren't on there before since Huawei has been on there for like years. But this like further isolates China from the US Semiconductor equipment and technology. They really can't buy anything now. And, and people start worrying about Nvidia's future revenues in China because of these controls. As market participants observe these trends, analysts begin to speculate about the long term feasibility of China specific products from Nvidia saying, hey, they have a workaround right now, but how long will this really stick around? There's a bunch of different ways to create limits on chips. You can just do total flops or flops per dollar, or, you know, Jensen in his Blackwell presentation said, you know, we're no longer on chip scaling, we're on energy scaling, which is like the most foundational level of computation. So, like, not even flops per dollar, but flops per watt is like, what really matters. And so you could just limit that. And, and then, and then, but then there's a whole question about, like, how do you measure that? What. What are you training on? And, and there's ways to get around that. There's always, like, loopholes. But you could. And we're seeing this now that, like, the, the, the definition of what's banned is growing and growing and growing. And there clearly are a bunch of ways to basically just tell Nvidia no. And that's kind of what's happening now. There's a huge debate over whether or not these chip sanctions should be in place and whether or not they should. They're effective. Right. So we'll get into that debate, but let's continue with the timeline. 2023. So China is facing persistent export controls, and also there's a global chip shortage, remember from COVID and whatnot. And also just increased demand for AI chips. So they continue to innovate with tailored products specifically for the Chinese market. Like the H800. Um, this has reduced performance parameters. Yeah.

Sounds bad when you say it like that.

It complies with export restrictions. At the same time, like every company, whenever there's, like, it's so easy to. If there wasn't the crazy geopolitical thing here, it's so easy to be like, oh, like the, like, you know, the, the speed limit is 65 miles an hour, so I'm going to go 64. Right. It's like every, every business is like.

Oh, I agree to shareholders.

Yeah. Like every company is like, oh, I'm like, like, I can't sell this product in this market. Okay. I just need to find a different product.

This is why over the last couple of weeks, if you're an American hedge fund manager and you're unclear of what's happening in the American market, even if you're a true patriot, you're going to look over at European defense companies and say, you know, this suddenly is interesting. Yeah, because they're just like, their mandate is to chase, yield and put up numbers for their LPs. And so that's just a unfortunate reality when it comes to these sort of geographic geopolitical sort of like hot button issues.

And so Even though the H800 is lower performance than the H100 which is the popular one in America, Chinese cloud adopters or cloud providers are just buying them nonstop because they need advanced hardware and they're willing to work around the limitations there. And so at this time, Ben Thompson writes Nvidia on Tuesday, this was yesterday said it would take a $5.5 billion charge after the US government limited exports of its H20 chip to China. And that's like kind of the main story that's driving the news today. So in 2023 Chinese tech giants accelerate their own product developments and this is what we're, this is like the hot topic in semi analysis right now. Huawei has this Cloud Matrix 384 and this and this. The goal here is to challenge Nvidia's dominance. And so it's described by the South China Morning Post as a nuclear level product. Sick. It's like Stargate but for China. Basically like they're using, I wonder if.

The South China Morning Post has any ties to the state.

I think they're just pumping up their bro. Who knows? They're just. No, no, obviously they're, it can be both. They're biased but, but you know they're, they're talking their book, they're excited about this. So the cloud matrix, it's funny because I had never heard like the, I mean I've heard like the nuclear analogy once or twice but Aaron Ginn yesterday wrote a whole piece about like this idea of like a nuclear cyber nuke in the form of a massive data center. So the Cloud Matrix 384 usually use it utilizes 384 Ascend 910C chips to deliver up to three 300 petaflops. So despite outperforming Nvidia's NVL 72 system on raw compute 300 petaflops versus 180 petaflops, the system is significantly more power hungry and costly, thus highlighting trade offs imposed by export controls.

So they're like great, we'll fire up the three gorges.

Damn. Yeah, I mean like they are adding 20% energy every year now in China and we're adding zero. So you play that out and being power hungry and costly, totally the trade you want to make just to be able to actually run the data center. Of course there is an economic calculation if, if, if they have, you know like an ASI system that's less economically efficient. It's like having you know, $100 drone. Shoot down a million dollar drone. You can only do that so many times. And so if you're running an AI system, it's fighting with another AI system. Oops, lost my headset. If you have one AI system that's fighting with another AI system, like at a certain point you need to be.

Able to, you need to be efficient.

You need to be efficient. Yeah.

This is the issue we talk about with, with US military countering drones with missiles.

Yes.

Like okay, if your missile costs 500k.

Yeah, yeah.

And you're taking out a drone that costs 600 bucks. Like just the math.

Yeah, that was the whole math. Behind the Roadrunner, you would just like.

Continuously launch them to just sort of like bankrupt apparently.

I think, I think a Patriot missile, which is an anti missile missile, shoots down a cruise missile, costs like a million dollars and it's not super reliable. So we fire to every single time and they don't come back. Like, they're just gone. Like if you, if you sense a threat, 2 million bucks. And so the Roadrunner, which I think is cheaper, I don't know the pricing but like it can fly and you only need one and if it doesn't, if it misses or there's a false alarm, it comes back and lands. Right. And so in theory, like this, economic warfare is important, but China has a huge advantage on power and energy. So the, and just to set the table a little bit on this, like what these systems are the Cloud Matrix 384, Nvidia's NVL 72. Basically these are a bunch of Nvidia chips that are racked together. And NVL means an nvlink, I believe, which is a system that links the different chips together so they can communicate kind of as one server. And then of course you do network them together, but allows you to store even. It allows you to treat the server as like a single abstraction, I believe. And so the main, the main difference again, power efficiency. And this is what Jensen has been talking about with Blackwell. He's really focused on energy efficiency, power efficiency, not pure petaflop output because this is becoming more and more of an economic calculation. So 2024. Last year, with further regulatory adjustments, 2024 marked the shift where US policy began targeting AI inference chips as much as training chips, prompting a pivot to inference optimized design. As the restrictions on training chips tightened, US Authorities announced new export control parameters set to become effective January 2025 that also scrutinized inference performance metrics. In anticipation, Nvidia began Developing a range of China specific GPUs for inference. And this is the H20, the L20 and the L2, each tailored to slight to meet slightly different performance and regulatory thresholds. And so Ben Thompson Yesterday wrote the H20 is a pretty lousy AI accelerator compared to the H100. It has fewer cores, much lower memory bandwidth, no NV link support. And this underscored the workaround that the workaround chips were compromises rather than true performance products. But it doesn't really matter because you need to have the best. Like if you're just thinking about the Chinese chip market, as long as it doesn't matter that you're worse than American Nvidia chips if you're still the best in the market. And so Nvidia can go, still win in theory until these were banned and make $5 billion. Yeah, but not anymore.

Or high flyer slash Deepseek is saying, okay, we're going to basically train our model on your model that you used all your H1 hundreds to make.

And, and the inference is the inference in theory, if you can, if you can, if you can do all these crazy workarounds to get the training done and you know, maybe export a bunch of GPT weights and stuff and do whatever you can to get the training done. The inference is where the economic value is created. Like you can think about training as CapEx and then the, the, the inference is the OPEX that allows you to actually derive economic value from the intelligence, the magical intelligence from the sky. Right? Yeah, like no one cares if there's just a super intelligent model that you can't actually inference. And so having good enough models, having state of the art models, open source models, but then being able to pump them out in a way that all 1 billion Chinese people can use them on a daily basis for all sorts of things to create new businesses, optimize their current businesses. There's a million things that you do with AI that should lead to GDP growth, that's really important. And so inference is becoming increasingly valuable. And so the phrase I was looking for that Jensen was using when he talked about the power efficiency metrics of Blackwell, it was ISO power. And so he's, he's really focusing on the, the scaling laws as they apply to just energy specifically not chips.

And this is the advantage that Nvidia has had by always being ahead is they have the luxury of yes, they need to continue to innovate, they need to come out with Blackwell the chip beyond Blackwell, but at the same time they're able to focus on these things that a second, third, fourth, fifth mover or somebody that's, you know, not in the same competitive realm doesn't have the luxury of thinking about energy efficiency.

Yep. And so we're, we're seeing like a bifurcation in the chip market and just the approach to design. So Jensen Huang is claiming that modern data centers, probably just in America are power constrained. ISO power means delivering maximal compute within a fixed power envelope. Newer generations like Blackwell greatly improve efficiency when you're under that 1 megawatt constraint, which is kind of where most of the, I mean, I guess a lot of data centers that want to do inference are sitting. Contrast that with Huawei's Cloud Matrix 384. This sacrifices power efficiency in exchange for higher aggregate compute. Semi analysis noted that while its raw performance is impressive, the trade offs in power usage make it unattractive outside of China where competitive pressures and cost structures differ. So if you don't have energy too cheap to meter, well then you got to get, you got to go ISO power and deliver maximal compute within a fixed power envelope. So as there's the preparation for the regulatory changeover in 2024, the new export control solidify Nvidia and other firms accelerate development of their next generation products aimed at circumventing increasing restrictions while meeting customer demands. So they developed the H20. Chinese companies, including domestic alternatives such as Huawei's in house designs and Alibaba's T head initiative continue R and D efforts to eventually replace place dependence on US origin chips. Because they see the writing on the wall, they're not going to be getting any US chips in the near future. So they've got to bring it in house. So analysts are forecasting that by 2026 AI inference compute demand could outstrip training demand by a factor of four, increasing the strategic importance of these new designs. Remember, just a few years ago it was all about inference because OpenAI would spend $100 million on a training run and then be like oh, like it's a toy or like oh, like some people are playing around with it or someone built a wrapper that's like okay. Now it's like everyone needs to inference constantly. And so as Trump comes in 2025, US tariff and export controls tightened to unprecedented levels. Obviously we talked about this a lot, raising rates to over 100%. There's some internal debate, but the new policies reaffirm a long standing US strategy to limit China's access to advanced semiconductors and semiconductor analysis strategies. In April 2025, the US government informed Nvidia that its H20AI accelerator would require an export license for shipments to China, effectively halting and triggering a planned $5.5 billion write off. Triggering a planned $5.5B write off. So Nvidia is expecting, they're kind of messaging this to their shareholders ahead of earnings. Taking the, taking the, the, the, the hit on the shoulder, right?

Nvidia is down 10% today.

10% because that, I mean I think 10% of their business is probably in China still and like now that doesn't exist. And sure, they'll probably make some money selling to Malaysia which will create a data center that China can train on.

And all this just comes down to.

It will be material.

This comes down to how somebody feels about these type of restrictions on China is going to come down to whether or not they truly believe that we are in an AI war with China. Right. We've been in a trade war, that much is obvious. We've had numerous cybersecurity incidents with China. We're clearly in this sort of economic war. And I think that Nvidia Jensen, Nvidia shareholders are probably super frustrated with this. But it's very possible that this is, you know, 100% the right decision given how fast AI is accelerating. And you know, even going back to the conversation with Aaron, you know, yesterday around the US needing to take this issue as seriously as we possibly can.

And so the dividing line here is whether chip controls are good or not. And Ben Thompson has been against the chip controls and I'll kind of read from his position. He says my head, my heterodox position on this matter is clear for the same reason that I am against the chip controls but in favor of banning the sale of semiconductor manufacturing equipment into China. I think that, that this is a good thing, that The Cloud Matrix 384 is dependent on TSMC and Samsung because that is a reason for China to maintain the status quo with Taiwan in particular. Semianalysis is much more hawkish on chip controls than I am. At the same time, the reality is that right now we are in the messy middle. We're not actually stopping Huawei from building a system that is capable of doing large language model training, albeit inefficiently. But we are hurting the fortunes of us AI champions of Nvidia by and limiting their long term competitiveness. And of course we are incentivizing everyone in China, from the government to private enterprise, to ultimately remove the point of leverage that we can't even wield properly. And so he's is. So Dylan Patel had a different flip on, different take on this. Let me see if I can pull it up.

He says we might need to have a newsletter goat debate.

We should, we should. I'm going to, I'm going to invite them both. But he basically said stop focusing on the chips. You have to focus on the entire supply chain. And as long as there aren't expert controls on everything in the entire chain, it's mostly ineffective. And so what else? Let's see. I think we covered most of this cloud matrix stuff. I mean, you can go read the full deep dive on semi analysis. It's fantastic and extremely deep as usual. But let's move on to the other story with Mark Zuckerberg. Yeah, he's. But first, if you're selling chips all over America, you're going to need to pay sales tax.

That's right.

Get on Numeral. Go to numeralhq.com. spend less than 5 minutes per month on sales tax compliance.

Numero works with a thousand plus different companies, helping them with sales tax compliance and automation. And we weren't working with Numero when we got this sample ordered, but we got a place for them right here.

Right here. Love to see it.

Thank you to Numerol for supporting the show. And go to numeralhq.com to check it out.

So Adam Carlson posted some screenshots from a story in the Wall Street Journal. But basically it's about how Mark Zuckerberg spent millions trying to suck up to Trump. So Meta only had to pay 450 million instead of 30 billion that the FTC wanted to settle an antitrust case against them. This is of course about the Instagram and WhatsApp acquisitions and the accusations that Meta is running a monopoly. And so Meta has offered $450 million to settle the antitrust case. This is far less than the FTC's $30 billion demand. And Zuckerberg has cultivated ties with Trump. FTC sued Meta in 2020, accusing it of stifling competition by acquiring startups. And it is crazy. I mean, this feels like tech history, like Instagram and WhatsApp that happened in like what, 2012, 2015, like that era. And we're now 10 years later and we're still litigating this. It is pretty wild. But we went through some of the.

Screenshots at the time. At the time of the Instagram acquisition, it was a photo sharing app that was not monetizing, Right? Maybe. Did they have any monetization at the time? I don't think they did anything else. So it was impossible to know at the time that this was going to be a historically significant acquisition. And it's very possible that there's no other company in the world that would have been able to unlock the value of Instagram in the way that Meta has. Yeah, right. And so it's not just that like.

Pundits and analysts at the time said, like, this is a massive waste of money. Like why is Zuck spending a billion dollars on Instagram? This is such a big acquisition, why he should just build it himself? Like, it doesn't, it's not necessary.

And back then, social media apps didn't just immediately copy all their most successful features.

Right.

It was like people kind of like stayed in their, in their lane a little bit more.

Yep.

You went to YouTube for long form. You went, you had, you know, sort of short form iterations on, over on snap on stories. And so everything has converged dramatically since then. And anyways, all the drama around this is just funny in the context of that we can't force, we have not been able to force the sale of TikTok despite it being directly against America's national security interests.

So the initial offer was 30 billion. Meta wanted to pay half a billion. I guess FTC chairman Andrew Ferguson found the offer not credible and was ready, wasn't ready to settle for anything less than 18 billion and a consent decree. As the trial approached, Meta upped its offer to close to 1 billion and it wasn't enough. On Monday, the trial kicked off. The FTC called Zuckerberg, who privately expressed reluctance about taking the stand to testify for four hours. That's brutal. He had to throw on the suit. He was not you post.

Can we pull up that picture that John posted this morning?

I mean he looks fantastic in the, in the Wall Street Journal photo but he's not able to wear his. His dripped out designer T shirt. He no chain today. It's, it's very sad when, when a tech CEO has to put on a suit, it's always a sign that things aren't going well.

But I mean you can tell the jiu jitsu is working. You can tell that he's hitting the weight.

He looks extremely strong.

His neck is going to clearly if he Keeps Growing at 1% daily will eventually be as wide as his shoulders are.

Yes, yes, yes. But I mean he is looking in fighting form because he is fighting all the time fighting in the FTC courts and also fighting in the ring or the octagon or whatever he fights in I don't know how it works. Zuckerberg was back on the witness stand Tuesday, where he faced questioning from an FTC lawyer over whether Facebook had paid $1 billion to buy Instagram to neutralize a competitor. Asked if he would have preferred that Facebook's own camera app would have grown faster, Zuckerberg responded, I guess so. Yeah. A billion dollars is very expensive. Love it. Former FTC chair Lina Khan told the Journal that the company's 450 million settlement offer was delusional. Mark bought his way out of competing, so I'm not surprised that he thinks he can buy his way out of law enforcement too, said Khan, who is nominated by former President Joe Biden. His proposed remedy, like his marketing strategy, like his market strategy, is let my illegal monopoly keep monopolizing. So Lina Kahn's coming for Zucker.

What's Lena doing these days?

Free agent associate at a multi billion dollar, multi stage venture firm. I could imagine that being a good, good spot to move on to clip some fees, earn some carry. What's not to like? If she can identify a monopoly with the best of them, she's going to be a great vc.

That's right.

Think about it. Just flip it over, start printing and so we can go through some of Ben Thompson's analysis here. I liked his framing. He describes the three Facebook eras. He is very against this case. He says he thinks it's a poor one and the government's market definition in which Meta's only competitors are Snap and me. We I've never even heard of me.

Wee what is mewe.com the next gen.

Social network we have to buy shares.

In me I guess if it's competitive with Meta, one of the greatest businesses of all time. If it's in that must be a fantastic business.

I now I feel bad because I feel like there's probably an entrepreneur who's working really hard behind me. You're welcome to come on the show. We'd love to learn him, let's have him on. But no offense, but not typically in the conversation with Meta. But good luck out there. I hope you're printing. Probably doing very well if you're being mentioned by the FTC is absurd, says Ben Thompson. Unsurprisingly, I find Meta's opening statement presentation, which is particularly focused on the market definition question, very persuasive. What is striking to me, however, is the temporal aspect of this case makes everything a bit of a muddle. To my mind, there are three distinct periods at play. Facebook in the mobile transition. That's 2004 to 2013. Facebook started in the browser. Only really became Facebook in 2016 with the invention of the newsfeed. And then it was mobile that forced Facebook's hand. Mobile made Facebook more just an app, which was good news for the company's fortunes. The dominance of the iPhone and Android left Facebook no choice but to become an ad company to give up on its platform ambitions. Remember the whole, like, sign in with Facebook, Facebook's gonna be everything. Unified messaging, like all of that kind.

Of needed to take back step. You go back to aol.

Yep.

That was America Online.

Yeah.

They had platform.

Yeah, yeah, totally.

And so that. That seemed like probably the best thing you could do if you wanted to create a lot of value.

Yep. And so then ads are a great business then. He defines the second era as Facebook's monopoly period, 2014-2019. Facebook's most dominant period coincided with the worldwide growth of the smartphone market, which was, for the company, a worldwide expansion in its addressable market, along with a worldwide expansion in its advertising base, which was predominantly mobile apps. This is when Facebook most definitely, most definitively won the social networking competition. Probably the peak of the company's dominance was in 2016.

This is interesting. So in August of that year, he wrote the Audacity of Copying well about why Facebook's plan to compete with Snap, which is supposedly the behavior the FTC wants to encourage, was likely to be successful.

Yes.

And so they were like, yeah, we.

Want competition, we want competition. You're going to compete with Snap directly in stories and all the different features, and then eventually reels. And yet. And yet it's not enough competition.

Yeah. And so this, he says the. The reason slots in very nicely with FTC's current case. This is why it is so fascinating that Facebook is leveraging Instagram in this way. For all of Snapchat's explosive growth, Instagram is still more than double the size, with far more penetration across multiple demographics and international users. Rather than launch a Stories app without the network that is the most fundamental feature of any app built on sharing, Facebook is leveraging one of their most valuable assets. Instagram's $500 million. Sorry, 500 million users. And so Ben says it's not a surprise that the initial version of this case was filed in 2020, just after this era ended. Unsurprisingly, the FTC hadn't yet caught on to this reality. The initial version of this case didn't even include the name TikTok, which is interesting, which is why the government's arguments are rooted in the assumption that personal and social networking is 1 the only pertinent market and 2 defined to be whatever it is that meta's apps do. So Ben putting them in the true.

Zone so he closes out with Facebook's competitive convergence. This era is 2020 to the present 2025. By 2015 I was already writing that Facebook needed to move beyond mere social networking from Facebook in the future feed and this is the video slop era where everyone everything converges on the identical feed of vertical video. Algorithmic content, comments, like shares we saw.

The following list doesn't matter anymore. It's just about the app serving you what it thinks you're most likely to keep using the app.

And there's a slide in here what people want to do today consume video and that's why we're doing this. Well, we have a an early investor in Facebook coming on the show, Ali Partovi. He's announcing Neo and a bunch of fun stuff, but it'll be interesting. Maybe we'll get his take on what's going on with Facebook since he's been in the company and been involved in Silicon Valley and angel investing for four decades now. But until he joins the studio. Let's tell you about Eight Sleep. They got a five year warranty, a 30 night risk free trial free returns, free shipping and Pod 4 has all the signature features you love about the Pod, plus new groundbreaking updates. It's. It's a fantastic product. Unless you're up late negotiating and then you get a terrible sleep.

Score.

I got a 54 last night. It's awful.

Anyway, brutal night. Let's see how I did. I actually. Oh, I got 100.

Oh, you got 100. Wow. Smoking proof of work.

Eight hours and 13 minutes slept. Oh well, two and a half hours of REM sleep.

That feels like a lot.

Which is a lot.

That's great. Is that.

You know why it's a lot? Because autopilot made adjustments to the temperature of my bed to boost it.

Boosted.

I got boosted.

You got boosted?

I got boosted.

Okay, well in the meantime, while we're waiting for Ali to join, I thought this post was funny. Pratik Joshi says someone accidentally referred to Eric Gliman as Eric Ramp Man. And as a founder, that's how synonymous you want to be with your startup. I thought that was very funny. And Juan says, my fearless leader knows no bounds. Synonymous with saving businesses time and money. He is Eric Boss King Time Money Saving Man. His last name rolls right off the tongue.

Ramp Man. Yeah, this is Eric it's not too late. You can actually change your name.

Yeah.

Whatever you're born with, it's possible to change it.

Ramp, man.

And if you're going to change, if you're going to go through the hassle of changing it, at least make it an ad.

This is the, this is the concept of, like the method entrepreneur. Like, live your brand. I mean, you're doing it right now with this. You're living the TBPN brand by wearing all of our sponsors on your jacket.

Not all. We're missing a few.

Oh, yeah, yeah. Most of our sponsors, they're getting on.

The next round.

Anyway. Did you see that? Someone is selling A Mint Fitbit Inspire 3 on the Bloomberg Terminal for $66. Special Situation says someone got margin called so badly last week that they have listed a mint condition Fitbit Inspire on Bloomberg Marketplace. The terminal itself costs like $3,000 per month. It's rough.

I just got to say we got to get a Bloomberg terminal at the new office.

100%. And I definitely want to. I mean, this seems like just, just this access. People are selling cars on here. There's a Cayenne Turbo V8 for $57,000. They're selling all sorts of stuff. VW Golf is on here for 6,000. Porcelain jar. 200 years at City for $100. Some of these seem like great pickups. Nespresso.

Somebody here saying, brand new Rolex Submariner date. Skip the Q. Put that thing on bezel.

Put that thing on bezel. Your bezel concierge is available to help you source any watch on the planet. Why stick to what's just on the Bloomberg terminal? Head over to Bezel. Find exactly what you want. Create a, Create a wish list. Talk to the concierge. Get the perfect watch. Don't leave it to chance on the Bloomberg terminal. Unless you're trying to get a Lexus LX 5702009 model year for 1975. Very funny. Anyway, I always find these, these, these, these screenshots from Bloomberg, like sales are always so funny because it'll always be like, oh, the market's down, somebody's selling their private jet or something.

It's great, but it's a good meme. But it's actually going on all the time.

Yeah.

Regardless of market conditions. Anyway, did you see this post? Somebody in the admin said the golden age of American innovation.

Not just somebody in the admin. It's Michael Kratzios.

Michael Kratzios. Sorry, I didn't even look at his name. He said the golden age of American innovation is on our horizon if we choose it. Today, we share the Trump administration's technology policy agenda. And then if you look in the agenda, it's got a bunch of. It says stagnation was a choice. We have weighed down our builders and innovators. The well intentioned regulatory regime of the 1970s became an ever tightening ratchet, first hampering America's ability to become a net energy exporter and then making it harder and harder to build. We seem to have lost focus and vision, to have lowered our sights and let systems and structures and bureaucracies muddle us along. But we are capable of so much more. Our technologies permit us to manipulate time and space. They leave distance annihilated, cause things to grow and, and improve productivity. And somebody highlighted that this line. Our technologies permit us to manipulate time and space, which sounds really dramatic. I think he's potentially just referencing.

I think Kratzios has discovered time travel.

Yeah, very possible.

He was formerly the CTO of America. He went away for a couple years, worked at Scale AI, maybe while he was there. He's tinking on time travel. You never know.

You never know.

Teleportation.

You never know.

We're so bad.

He's got a clearance.

Anyway, we have Ali in the studio. Welcome to the show. How you doing?

Hey guys, do you hear me okay?

Yeah. You sound great. How are you doing?

Great to see you. I'm awesome. I'm so excited to be here. Love the show. Great to see you. John and Jodi, great to have you.

Can you kick us off with just a recap of the announcement? What's going on today? Give us a little bit of your background and then we'll go from there.

Yeah. I'm so excited. Today we just announced that we have raised $320 million in new funds.

Congratulations.

4. Thank you. It's our fourth fund on the heels of really successful bets on companies that we were the seed investor in, like Cursor, Ki, Blue sky and many more. And I'd say, more importantly, I'd say this is the moment where NIO is really coming into its own as becoming Silicon Valley's premier community for tech talent and startups, which is, you know, I've been a member of a lot of great tech communities and it's very exciting to be building what I think is the next really great source for innovation in Silicon Valley.

Yeah. Can you talk about some of your early career investments and how your strategy around investing at the earliest stages has evolved over the Last, I don't know if it's been 20 years of investing for you, maybe more. How long have you been doing this?

Well, I was a CS student at Harvard and started a company shortly after college with two other Harvard grads, a company called Link Exchange that was one of the first advertising networks on the Internet. And the success of that, when I was 26, we sold to Microsoft. Ever since then, I've been a mentor to other startup founders. And I realized that the most surprising and important lesson of my career was the difference that a single individual can make. And I realized there was a super skill I wanted to have, which was the ability to spot talent at a very young age. At the time, it was trying to hire the smartest young people to join my company. But then this kind of pivoted towards being a mentor to young founders and then an angel investor in those founders. And so I was fortunate enough to be really early in companies like Facebook, Dropbox, kind of Airbnb. I missed the very early round when Brian Chesky accidentally kind of pitched me in a restaurant. He was not even at my table. It was like a chance encounter. But I invested in, invested in a later round. So, yes, also Zappos, Uber, couple of others. And yeah, I'd say the common theme that was not just being an investor, but helping these companies build their early teams and impressing on them the importance of hiring exceptional people. And in 2017, I started NIO with the really audacious belief that it is possible to identify future tech leaders when they're still in college. And not only that, but that we could build a machine, a systematic institution that identifies superstars and helps them maximize their potential. And today I have to pinch myself that I really feel like we've proven that this is possible and that we're doing it systematically. NEO scholars have founded companies like Cursor, Pica, Cognition, Chai, Discovery, and many more. And at this point, it's an institutional system, not just like one person going around with a knack to make investments.

The meme of dropping out of college to start a startup has been fundamental to the history of Silicon Valley. Do you think more students should be dropping out? Is it not enough dropping out?

And how early is too early? Should they drop it out of high school, middle school?

Should they get a year in the dorms to get a little bit of.

I got a 4 year old. If you want to write a check, he's ready.

I love it. First of all, when you mentioned meme, I thought you were talking about there was A Twitter meme a month or two ago saying these are the institutions that students would drop out of Harvard and Stanford for. And it showed Neo, it showed the Teal Fellowship, but I loved it because it also showed Zack Frankel. So.

We don't say that word on the show. We don't say that the F word on the show. But he's a legend.

He's a one man institution. But to your question, first of all, if you have a four year old, have them use code.org, which is the nonprofit my brother and I founded that teaches young kids how to code. Literally. I started my 4 year old on it, highly recommend it. The question about dropping out of college, I think about it all the time. I mean some of the names I mentioned were dropouts. And one of our best investments of the past year was in Sapien, which was three dropouts who didn't even finish sophomore year at Harvard, Stanford and UT Austin respectively. Having said that, I think there's way too much of a emphasis on you have to drop out and you're not good enough if you don't drop out. Among Silicon Valley, obviously not everywhere. But in the tech industry, I think it can be a little unhealthy. What we offer is a gap semester grant. So it's like a Teal Fellowship, but much less scary because at the end of the semester you can go back to school. In fact, that's expected. It's just spend this, take a semester off, but build something of your own experience, entrepreneurship. And if you strike it big, keep going, you can then drop out. You can then and then would fund you and you know, there are some people for whom that would totally make sense. But if you know, but if that doesn't make sense, I would say it's better to stay in school until you have found the thing worthy of leaving school for the thing that your heart and your passion is really ready to commit to for years.

Yeah, this is kind of the story of Mark Zuckerberg. He takes the summer off and is actually on deferred referral and then eventually never goes back, but then gets the honorary degree years later. But, but I, I have to ask, I mean, I want to go into the present, obviously, but how did you get in the early Facebook rounds? Like, what was that story? I know Brian Chesky. You meet at a restaurant. Was it a different restaurant?

My stories as an investor are mostly about what an idiot I am. And I'm really, honestly, I'm not being humble. Brian Chesky. I'm at a meeting with a Wall Street Journal reporter that I thought was an important meeting. Random person at the table next to me pops in and starts entering our conversation. I think the Wall Street Journal guy actually started asked him a question and then before I know it, like this guy's pitching me his, you know, his thing, Airbnb. And I remember thinking this is such a cool concept. It's batshit. It's super ambitious, but if it works, this would be huge. But I was also annoyed. Like, don't you see, I'm in an important meeting with a Wall Street Journal reporter here. So I never forgot him. But I didn't, didn't follow up. To my, to my regret, I invested in a later round Facebook. I cannot claim any credit for my twin brother. Hadi, who is absolutely brilliant, deserves all the credit for that connection. He was introduced to them by Sean Parker when it was a nine person startup and we referred the very first summer intern to Facebook, Darian Shirazi and served as advisors. But at the time I remember talking to Hadi and saying this company seems like an online fraternity. I'm not sure I see the business here. It seems like it's really cool and I wish I could be back in college and be a member of this. You had to be a college student to be able to use it. It had maybe a million users. And I remember Hadi saying to me, no Ali, we're not betting on the company, it's the person. And this 19 year old mark Zuckerberg reminds me more of Bill Gates than anyone I've ever met. And the team around him were stunning.

How do you think?

If it hadn't been for my brother, I would not have been involved in Facebook. But this was 2004 or 5 or so and 2005 and was the germ of this belief of investing in people that is the core of what NIO is today. And Neo today we discover college students long before they start startups and start mentoring them. We invest in them by investing our time.

Yeah.

Do you think it is has been very like my friends that are investors. I think it's actually helpful if they get a huge win early in their investing career because they one know what, what great really looks like. They know what great operators look like and they have less FOMO in the st sense that you know, when they get pitched an idea that they like but the founders maybe not the right fit, it's just easier for them to pass. Do you think, do you find that, do you find that like you know, getting, you know, big early wins has allowed you to like, you know, be.

A better risk on even.

Yeah, yeah, more, more risk on. But also just being okay, not like having that sense of FOMO being like, sure, I like this idea, but you know, I'm not going to invest.

Yeah, I think, yeah.

I'd say I still have all sorts of insecurities, both the FOMO as well as, you know, sometimes being unsure about what something is, the right investment. But what I've definitely got incredible confidence about is my ability to spot talent and, and you know, when I started nio, this idea that you could find future tech leaders when they're still in college, this was a kind of a crazy suggestion. Now I feel incredibly sure about not just my ability, but the company's ability to do that. But the other thing I'd say for me, I would say my path as an investor has grown more from my mistakes than from those early wins. I remember when the smartest engineer at my own startup left to start something. He was a contractor. This guy Max Levchin, who was 22 at the time, started what was called confinity, which became PayPal. We all knew Max is brilliant, brilliant as an engineer. And we also all thought that Confinity was kind of a joke. It was like an app for the PalmPilot to send money with infrared. And today I would just say if Max is involved with anything, whatever involvement, I would invest in it because of the person. Whereas back then as a 26 year old, I had a lot of hubris. I just sold my company, I thought, I'm the bee's knees. And so when you're too arrogant as an investor, you start trying to pretend like you're the smartest person in the room and poke holes in the other person's idea and come up with ways it would fail. Whereas today my approach is maybe I'm not the smartest person in the room. Let's figure out if the other person is actually the smartest person, even if they're half my age. And if so, and if they're committing their life to something, I just want to support them and help them. And maybe I have wisdom to give or ways to help but not miss out in any way. I should say similar, miss Same time frame, 1999 or so. One of the smartest people from Harvard computer science, Craig Silverstein, kind of reached out to a whole group of us saying, hey, I need advice. I'm joining as the CTO of this startup. My Stanford PhD friends Larry and Sergey are starting to. And I remember thinking Poor Craig, he used to be so smart. Doesn't he know there's 10 other search engines, this Google company, just because they have a better algorithm, you know, and honestly, it was a foolish business idea. Like if you were to apply kind of MBA type, is this a good business plan? You'd be like, it wouldn't make sense. What made sense is that the people were freaking brilliant. And today I would in a heartbeat bet on people of that caliber, even if they're taking on much bigger companies.

Those are besides raw intelligence. What are you looking for?

Yeah, what's a conversation with you? Like, if you're a sophomore in college, what are you looking for?

Well, I mean, the one to talk about is Michael Trul, the CEO now the CEO of Cursor. I met him when he was a rising sophomore between freshman and and sophomore year in college when I was interviewing him to become a NEO Scholar. And so at the time, today to become a NEO Scholars, thousands of people apply through our website. But back then I was basically the website. I was like, you know, it was such a grind visiting different campuses and so on. And I would give people technical interviews instead of what's today an online coding test. So I gave him a coding test equivalent to what? It would evaluate whether he could get a job at a big tech firm or so on. He crushed it within the first 10 or 12 minutes, which meant we had the rest of the hour. I let him give me a coding test, which was actually much harder than the test I had given him, and I solved it. But that ate up much more than the other one. But the real thing I look for, besides I'd say technical ability is a kind of. It's an important prerequisite. We back technical founders. But I asked myself, how magnetic is this person? The word magnetic, specifically, I'm asking myself, if this person started a company, how many of his or her smartest friends would drop whatever they're doing to join it? And you can get a sense of that just from your, from the vibe of a person. You can also get a sense of that from looking at what else they've done. Michael Trull had had two startups when he was in high school. I wouldn't call them startups, but they were like small businesses that he had run. And there's other people, you can see where you can see what their role is in campus organizations. Do they join campus organizations or do they become the president of the campus organization or do they start a new organization? And yeah, so how magnetic are they? And then I'D say the other thing is willingness to kind of not accept the rules or not accept the status quo. I'm not talking about breaking rules or breaking laws, but just to see beyond them and to have such a kind of dogged commitment to what you're doing that if you see a rule or a law or people saying, this is never been done before, never been done before, to figure out a way around it, including changing the law or getting the, you know, getting the government approval. So a company like Ki. We were the. One of the earliest backers in this is. I guess it's a tech company. But I'd say their biggest innovation is regulatory. They have legalized trading on event outcomes, starting with things like pandemics and Covid to election results. And now they're. They're now the largest sports within two months or three months of launch. They're the single largest sports betting market in the United States, all legally. When everyone told them, this is not going to be legal, you'll never succeed at this. We bet on them because of their doggedness to kind of see beyond those naysayers.

Jordy, how much time do you spend thinking about the future today versus just obsessing over finding talent? Right. There's a lot of venture capital firms that spend a lot of time thinking about their visions for the future, you know, understanding markets. Right. And that's really fun to do. And it's a strategy that can work. And when it does work, it's like, amazing, right? We all love these stories where VC puts out like a request for startups and then they get, you know, and then somebody kind of like goes and does it. Like, it's, it's amazing. But at the same time, like, to me, you probably, I would imagine you're spending time thinking about the future in the context of what an incredible founder's vision for it is.

Yeah, it's a great question. Look, I love business ideas. In fact, you know, I am starting a new business on my own on the side with that. I've got an amazing team that we're incubating now. And I love brainstorming business ideas, but I don't spend time making predictions about the future and kind of claiming that I know what's going to happen. Frankly, I just. Maybe it's insecurity or humility, but putting out a request for startups where we say, this is what's going to succeed, that's not how we operate. We focus on, is the other person smarter than me? We're obsessed with finding the Top talent. In fact, I'm, I'm dialing in from Stanford University. I don't know if you can see the background right now. You know, like, yeah, I'm in a room in the room at the Huang Engineering Center.

You, you, you drop out. Drop out right now. Come over here. Come over here. Take my money.

You bring check?

Yeah, yeah. He's flipping a sign on the. On campus, but it's just a big check. Just say yes, drop out. Drop out. Please take my money.

Joking aside, what we do actually is we come in. And not just at Stanford, also at Waterloo, Harvard, Princeton, Yale, mit, Myself and my team members will go in and provide an entire day of office hours. If anybody wants mentorship, book 20 minutes or 30 minutes. And it could be on your career, it could be on job search, it could be on some startup idea you're thinking of. We almost never try to encourage anyone to drop out. We're here to lead with mentorship and helping young people. And by the way, I don't even define Neo as being a VC firm. If you look at our website, we call ourselves first and foremost a mentorship community. Our north stars are at mentorship and community. And our business model is investing. But I can tell you, it is so exciting to meet a college student who might be 19 or 20 and to realize I'm talking to somebody smarter than myself, whose potential is to be bigger than myself. And, you know, that's what I felt when I met Michael Trull five years ago and started investing in him. And what I mean by investing means spending time, inviting him to events, you know, inviting him, connecting him to my network. We introduced him to his first startup internship at a company called Octant. His MIT friend Aman Sanger, we connected to his first startup internship at a company called u dot com. And when the two of them, two and a half years later their senior year, said they're thinking of doing a startup, I started booking time to mentor them every few weeks to help brainstorm ideas. When they decided finally that they're ready to fully pursue it and they weren't dropping out, they graduated. We were like, no question, first to invest in them.

And.

At the time it was just two guys and an idea, and they even changed the idea. So we were just investing in the two guys.

I'm curious, how many times a year do you end up having one of these founders saying, like, hey, can you talk to my parents? They want to know if the story of Michael Jordan. Yeah, yeah. When they were into the family, you want like, total Alignment between, you know, the entrepreneur and the family. Like, ideally it's not necessary. But I'm curious, I'm sure you've had some of those conversations being like, no, I'm going to give your son a million dollars. And yes, you know, they're not going to have a degree, but like, you know, they have a real, you know, I think not every parent understands that, you know, just getting across the line and raising venture capital is a signal to the market. That's similar quality in many ways to a college degree.

What I love about this question is that you recognize the human part of what's going on. This is not just like, oh, how much money? Or so on. It's not just about startups. We're talking about human beings who have the same hopes and fears and parents as you and I do. And I love that you are in like 20 minutes thinking of the things that it's taken us a few years to think of. But yes, I speak to parents every year. A few, a few one on one type zooms. But this year we actually created an event called Parents Weekend where we flew out the parents of the Neo Scholars all for like a weekend long of like talks by Scott Wu from Cognition and other, you know, other people who, whether they dropped out or started a company a couple of years after college leaders that the parents could look to and see, oh my gosh, yeah, I guess my daughter or son could end up being like this person. And this is legit. And it's, you know, it's not riskier or less prestigious than getting a job at, you know, OpenAI or a ramp or whatever big company.

Can you talk a little bit about the business model right now? I mean, if you take a, a scholar who is going to do something entrepreneurial for the summer, is that like we're setting up a C corp for you, we're taking 7% or something like that. And then if it winds down at the end of the summer, you go back to school, that's fine. But we're set up for the corporate journey and just putting you on Sand Hill Road and raising a mango seed. Or is it more like, you know, teal fellowships and nonprofit. But then obviously once you do the TILL fellowship, you go out and raise and you know, plenty of people aligned with the Thiel Fellowship have access to those deals, et cetera. But how do you think about just mechanically going through a partnership with you?

Yeah, so in some ways I would say we are the signal of finding the great undergraduate talent that the Thiel Fellowship Once was, you know, if you talk to anyone at top universities, they now what they'll tell you is you apply to become a Teal Fellow after you've dropped out and after you've raised 3 to 5 million dollars in funding, then become a Teal Fellow. And the majority of the Teal Fellows followed that path. They became a Teal Fellow long after raising money, long after dropping out of college. We identify people years before they might start a company and the main thing we offer them is not cash, it's the network and the mentorship and the real care and attention to helping them figure out what whatever path is best for them. That might be graduate school, it might be getting a job at OpenAI. Last year OpenAI hired all of their new grad hires through Neo and it might be a job at a startup. If they want to do though, like you said, do their own thing. The way we are currently structuring it is a gap semester grant. So it's not a summer, it's a gap semester because we want that person to do, to show a little sacrifice. If it's a summer program, it'll attract, you know, funtrepreneurs who are just kind of looking for a summer camp and oh, this will look good on my resume instead of interning at Google. You know, there, there might be other programs that do that, but our view is if we're going to give them a grant, we'll make them take a, we'll offer them a chance to take a semester off with other students doing the same thing in a shared workspace. So we provide shared space, space, community, so you're not alone. And $20,000 equity free. Just grant, grant and mentorship on a week to week basis. I'll spend time with them. We have other mentors who spend time with them. But this aspect of being in a shared space and having the other students who are doing the same thing is very different than programs like the Till Fellowship.

So of the 350 that you just raised, some of that's going to, some of that's going to grants and then some of that's going to probably participation in seed rounds once the C corp set up. And then, but then like are you trying to do pro rata, are you trying to concentrate into the best companies as they grow or will you set up a separate growth vehicle or is that just not in the cards?

We're not planning on doing a growth vehicle and it's, this has been, is a very conscious decision because it sounds like you're quite Familiar with how VC works. The standard path, if you're a VC and you're ambitious and you want to be number one, the standard path is to raise more and more and more capital, which means invest in later and later rounds and become, over time, an asset collector. It's, you know, you got to go public.

That's the new hot thing in 2025. If you're not taking your firm public this year, what are you doing exactly?

But, yeah, the industry term for it is AUM Bloat Assets under management Bloat. And we consciously thought about this and said, that's just not us. That's not what we want for our future. We would rather bet on our own returns. We would rather grow in quality than quantity. We'd rather stay small. So the program I mentioned for college students only accepts 20 to 30 Neo scholars a year. We also have an accelerator, which we haven't talked a lot about, but it is aiming to be the most selective program for young technical founders there is. And it offers 600k, you know, standard deal to participants. That's only funding 20 companies a year. We want to stay selective rather than ballooning and growing based on volume. And that said, we're also very ambitious. So we were thinking about how, what's our path for growth? What's our plan for world domination? And it's betting more on our own returns. And what that means is this fund, even though it's barely larger than our last fund, our partner's investment, meaning, you know, my own money from my pocket into this fund is far more than any of our previous funds. In fact, it's about the same as all the previous funds combined. And so we are going to be betting a larger and larger share of our own money into. Into our funds over time, and essentially betting on our own ability to. To identify killer talent.

And we love betting. We love betting. On yourself, your own fund. Congratulations on the fund. It's been fantastic talking to you. Thanks so much for coming.

I'm excited to see the new scholars and companies coming out of neo.

Yeah. And then snipe them and do the rounds before Ali can get to them.

Yeah. Just kidding.

Good luck.

Be wary. Be wary of, you know, you don't.

Want to publicize these guys too much.

If you get. If you get some spies in neo, they're going to start front running.

Front running.

You got to get your cyber security up.

Yeah, yeah, I. I am. I'm not worried about that because we have such good relation. I could tell you right now who the college students right now that are going to be future tech leaders.

I love.

And There are other VCs who look to see who Neo scholars are. In fact, the Forbes feature we had this morning, Christina Shen, who is an absolute Superstar, formerly from A16Z, now she has her own firm. She said Neo is the number one signal that she looks to more than any other program or accelerator for where future founders are. And yeah, so it is hard when we do this and identify someone so young because then there'll be actually competition for who gets to invest in them. But our view is that we want to be their most supportive advocate and mentor and earn the privilege to invest in the tech leaders of tomorrow.

Well, your work ethic is inspiring and the only thing I would ask from you is put as you leave that room, put TVPN on on the tv. On the TV behind you. Just let it play.

I'm going to put it on the whiteboard here right now.

There we go. All right. Thank you, Ollie.

Thank you so much. I love it.

All right, there we go.

That's good. There we go. Thank you. Have a great rest of your day.

Cheers.

Bye bye.

Next up we have we're pivoting back to China. There's a new IFP report from the Institute of Progress all about Nvidia and the H20 problem inference supercomputers and US export control gaps. We have Tim coming on the show to talk about it. Editor's note at the top of this thing, shortly after this piece was published, Nvidia disclosed that the US government had informed it that the export of H20 chips violates supercomputer end use restrictions as we claim in the piece of we applaud the Bureau of Industry and Security for its quick action. So yeah, they probably read this. There's probably a pre print going around, but it's a very, very detailed report all about chip competition and everything we just talked about in the opener and in the deep dive. So I'm excited to get him to break some of this down. I'll read from the summary while we're waiting to get set up with him. The United States is on the verge of repeating a critical strategic error. Despite having the legal authority and technical tools to constrain China's access to advanced AI chips, the US Government is failing to enforce existing export controls or adapt them to new threats. As a result, Chinese firms are exploiting loopholes to acquire powerful US hardware, undermining America's lead in frontier AI. The most immediate risk is the pending export of over 1.3 million Nvidia H20 chips worth more than 16 billion. When we talked to Jordan yesterday from ChinaTalk, he, he was saying it's $2 billion. Like that's a lot. And we were, remember I was saying like, that's not that much like Stargate's 500 billion. Well, it turns out that the proposed number was actually more like 16 billion. Jordan very thoughtfully posted and mentioned me to flag this, but it is a, it is a much bigger number than people thought. Of course the chips are worth 16 billion. The, the, the economic impact to Nvidia is 5.5 billion. And these chips were going to go to Chinese tech giants including ByteDance, Alibaba, Tencent, etc. They're optimized for AI inference. Odd. One second, we're getting some errors. The team will email you.

While we're getting errors. I'll just put this up.

Put this up. Yeah. I mean, it's a great time to do some, some ramp ads. Some ads, all types. What else do we got in here? We've done numeral, we've done eight sleep, but we haven't done wander. Find your happy place.

Find your happy place.

Book a wander with inspiring views, hotel grade amenities, dreamy beds, top tier cleaning, and 247 concierge service. It's a vacation home, but better folks. And we can go back to the ifp. So these chips are optimized for AI inference as we discussed, and they're likely to be used in Chinese supercomputers, which violates US Export control controls. At least one of the buyers, Tencent, has already installed H20s in a facility used to train a large model. Very likely in breach of existing controls, restricting the usage of chips in supercomputers exceeding certain thresholds. Deepseek's supercomputer used to train their V3 model is also likely in breach of the same restrictions. Can you. Should I just call him? Let me see. What would be the best.

He's having a zoom issue.

Yeah, he's getting an error on his end. Something. Let me see. Let me find.

You could try on his phone too, maybe.

Let me see.

All right, I'm going to keep reading. The failure is part of a broader problem. Problem? Broader pattern. Huawei stockpiled restricted chip components, including chip dies and high bandwidth memory before controls took effect. We talked about this earlier. Just stockpiling that happened. Over the last few years, TSMC has facilitated production of over 3 million cutting edge chip dies directly for Chinese firms for Chinese AI efforts, sometimes in violation of US chip rules. DeepSeek a Chinese lab, trained state of the art models using American chips sold legally due to delayed restrictions. The rationale behind export controls remains sound. AI models trained and deployed on the most advanced chips pose national security security risks. But the administration of export controls is failing to keep pace with new developments in the fields. Inference, once seen as secondary, has become central to training, fine tuning and increasing the capabilities of deployed models. The H20, a chip specialized in inference, is already 20% faster than the H100 for inference tasks. That's very significant.

And so this is where they get to like straight up policy recommendations. The US must act now. We recommend that BIS should immediately block the age 20 shipments using its existing authority. And so a lot of times these laws are written and people just don't really know how they can be interpreted. Obviously different companies are arguing for different interpretations. It lands in the courts. But IFP is arguing that there's an existing rule here, 15 CFR 744.23 that will allow them to block H20s, which I think is what happened. Nvidia must investigate and halt transactions that raise legal red flags under its know your customer obligations. BIS should update export controls to cover inference chips, which is what's happening. And the White House should empower a technical team, ideally within the AI Safety Institute, AISI at nist, to forecast AI threats and proactively shape control policy. And five, BIS should track chips once exported to high risk locations by incentivizing industry adoption of chip geolocation features.

Yeah, we've sort of referenced this before, but it'd be interesting to kind of get a sense of what the black market chip market looks like. Yeah, just because you can imagine chips flowing to various countries, you know, to presumably an end customer that then get resold and moved across borders, things like that. So very fascinating.

X is saying that zoom is down.

Oh really?

It seems like there might be actually like an attack.

Like I think there is a massive.

Spotify and Apple are having trouble today. But you know what, the haters of X, X seems to be working.

We're live, we thought we're live on X recording downages across the rest of the the Internet.

Yeah. And everyone was saying, oh, X is going to be the first one to go down. He can't possibly keep the servers on. That Elon guy knows nothing about software or engineering. But the X server stays live and I think we're still live on YouTube, which is great. Anyway, maybe we should do some timeline while we get that set up. Settle into this debate all about Katy Perry.

Yeah, we should. When I heard that Apple podcasts and Spotify were down, you know what I thought about? Big radio.

Big radio.

Big radio.

We really needed. You don't hear about big radio deal. You know, the satellites are still cranking out content. You just turn on the radio when the Internet's down in your car, turn it out. You get three years free when you buy a new car. And TVPN should have its own channel on there.

We should do a deep dive on Sirius xm.

Interesting business. Yeah, Yeah, I want to do more of those, like, weird niche deep dives that aren't covered that much. I was reading about Jim Cramer's career yesterday. That was really interesting one. Even the Johnny Carson show and how that, that, that, like whole industry developed. Like, have you followed any of the. Have you followed any of the late night wars? Are you familiar with this? So Johnny Carson started the late, Late, late night, or the Tonight Show. It's called the Tonight show with Johnny Carson. He started that, eventually became like the highest paid celebrity in all of Hollywood. And then when he retired, Jay Leno took over. I believe I'm getting this right. I'm kind of doing off the top of my head, but Jay Leno took over. Jay Leno had, like, another historic run. And the Tonight show is, like, more popular than ever. So Jay Leno buys the McLaren F1. He builds, like this massive garage of, like, hundreds of cars, has, like, he's fabulously wealthy. Then there's this weird. There's this weird wrinkle where he's going to retire. And I think NBC owns the whole thing. They say, hey, Jay, it's time for you to step down. You're under contract. We're going to maybe put you on something else. You're going to go to an earlier slot, I believe, different show, but which I think would be like, not a big deal at all. It didn't go well at all. So Conan comes in, takes the Tonight show, and. And that's, I think, kind of working. But there's always an adjustment because the audience is slightly different. The lead in is slightly different. It's like, because if you think about it, if you're just. The average TV viewer is like watching primetime. So they're watching CSI or they're watching Law and Order, or they're watching, I don't know, like, whatever is on primetime. I don't watch tv, but, you know, these normal shows.

Yeah, prime time, you know, prime time, yeah.

It's a Modern family. That's like a primetime show. Right. Those Types of things. So you're watching that and then it flip. Or the Simpsons, right? It probably comes on seven or whatever. Then that flips over to late night and you're getting the news, but in this like funny, comedy variety way with interviews with celebrities and stuff. But then the second, there's. There's actually several blocks. So there's like the 10am late night show, then there's 11, 11pm, 10pm, 11pm midnight, etc. And it kind of moves through. And so if you're the second show, you can't just repeat all the same things that are on the first show because then the person might be. But why am I getting repeated information? And so you kind of have to retool the show. Anyway, Conan comes in, Jay Leno doesn't like it and gets like. He's like, hey, I want to come out of retirement. I actually want my show back. And so he comes back and that kind of kicks Conan out of his whole flow because he had a show. And so then he goes off to like TBS and does his own show, starts a podcast. He winds up becoming very successful. But it's like this big drama about how that happened. And then I think the final Tonight show landed with Jimmy Fallon, I believe, one of those guys. There's so many of these late night shows, show hosts at this point. But anyway, we should dig into how.

That talks about Blue Origin.

Origin, yes.

Maria posted yesterday on X Ordyhaze Ohn Coogan, please discuss Katy Perry and Blue Origin on the pod tomorrow. Here we are discussing it. Do you want to give some context, John?

Yes. So when I first saw this, first.

Of all, Monday morning on every single mainstream television, every mainstream television outlet was reporting this live or relatively live.

Yes. We were in the gym. And every TV was, it was, as they say, wall to wall.

It was wall.

It was wall to wall. But it was very interesting watching because. So Gayle King was one of the crew members, right? And she is an American TV personality. On CBS Mornings, she co hosts CBS mornings. And so CBS was like, this is our girl. Like, we gotta cover this for an hour. And then you'd flip over to what NBC was doing. And NBC was like, this is important, but it's not our girl that's there. So, like, we're gonna cover it for 20 minutes. And then even CNBC and Fox Business would cover it for like a few minutes just to be like, hey, this thing just happened. It's live. So just like tune in for a little bit. The rocket launch, like give you A little update, but they're not like lingering on it. But anyway, I thought the take that people wanted to hear was just like space travel school or like space tourism is cool.

First of all, space tourism is cool.

Yeah.

It's cool to see these big celebrities going up and it not just being another, you know, guy who sold some company in the 90s and wants to go to space. Right.

You mean Jeff Bezos? He literally was on the first Jeff Bezos. But he did the whole thing.

The history of space tourism is usually.

Yeah.

A technology brother.

Yes.

Who just really wants to go to space, figures out a way to get it done. It's not as high profile.

I mean, that's the story of the new NASA administrator. The new NASA administrator did the first civilian spacewalk on SpaceX. Like he went to real space. So Blue Origin takes you past the Karman Line, which is like 300 nautical miles, I think, something like that. And so you go to space, but like just barely technically. And you're not. It's not nearly as risky or aggressive as being like on the International Space Station, like way, way up there.

And so what was the first of all one? Space is awesome.

Yep.

I thought they, they handled themselves very well.

Yeah.

Just a really cool story.

Yeah.

Bezos was looking diced.

Yes.

He did fall right next to the fall. I did see the fall, which is unfortunate.

But, but, but he picked himself right up. He's clearly in the gym. He doesn't skip leg day. Cuz he popped. Right.

Maybe he just wanted an opportunity to kind of push up for sure.

For sure.

Off the triceps. But what was the controversy surrounding it? So a lot of people came out and said it was faked or certain aspects of it were faked. Yeah, I'm not going to put on the tinfoil hat because I don't.

Well, I think that, I think that there's an alleged conspiracy that they never even left Earth, I guess, which is very funny. But I do think that there is a conspiracy here that we should get into. So this is a viral post that got 44,000 likes and 11 million views. Here's definitive proof that the Blue Origin mission was faked. And it all centers around the fact that when they landed, the door opened and Bezos is there with like one of the hosts or someone, and they kind of say like, hey, close that door. And then they wait and then they open the door and everyone comes out. And so there's this question about like, how did they open the pressurized door without like knowing. Right. And so. But the interesting thing. Oh, you just want to put the thing in here. Okay, cool. Just up here?

Yeah.

Hey, Tim, how you doing?

Not bad. How are you?

I'm good.

Are we in the midst of a major cyber attack?

Yeah. What's going on?

Spotify was down this morning. Apple was down, Zoom's down now, Remote workers are in shambles.

Yeah, it's a disaster. I mean, I guess everyone is just taking the day off.

I appreciate you guys have a backup plan. This is great.

Yeah, it's great.

Well, can you give me the high level on the report it seems like you published right before there were some changes. Give me the high level and how things are developing.

And in all, half joking, half serious, but do you deserves some credit for the shift here?

Yeah, it seems on the credit side, I don't think we can take much. It looks like there was a lot of things happening behind the scenes. I will say the fix that we have now is just a short term thing and there's a lot of long term stuff that needs to happen. But so I guess back up as to what's actually going on here. So, yeah, it's been a pretty whirlwind 24 hours for the AI chip export controls. So you guys might know, in early 2024, Nvidia launched this new AI chip, this AI GPU called the H20. It was specifically designed for the Chinese market. So it was designed to stick within the computational limits that were defined through export controls. But it turns out that the H20 actually had great performance for doing inference. So running AI models opposed to training it. And this resulted in a situation where actually it was really good at inference and even better than some chips that were already banned. And this all happened during this sort of 2024 trend of inference for AI models becoming way more important for capability growth through models like O1 and Deep Seq using test time compute as well as other techniques like synthetic data generation. Rl. So basically a bunch of things sold like crazy. Over a million were shipped in 2024. And so this is like a really big deal. We had inference compute flowing freely from the US to China during this period where inference compute was becoming the key strategic input to AI development. And people like myself were going, well, Washington should really be reacting to this. But they were really slow to react. And it became sort of a couple of things that we learned last week. One, there was a meeting between Trump and Jensen, the Nvidia CEO at Mar a Lago, which is apparently a $1 million ahead dinner. And after that dinner, the White House reportedly reversed course. I mean, he plans to control the H20 chip. And then last week we also learned that Chinese companies had placed massive new orders for these H20s. So over 1.3 million so far this year. And this is like recently over just the last week after this, a lot of calls, including from us, to stop these orders going through. And yesterday, yeah, it became clear that BAS is actually. BAS is the Bureau of Industry and Security who does export controls, has taken action on this. I just threw a lot of info at you, but that's just where we're at now. Could talk more about sort of what we think should happen next, but yeah, it's a pretty crazy situation at the moment.

I'm curious, what, what's your broad stance on. On Nvidia at the moment? Jensen has obviously been in a tough space of, you know, being. Being an American, yet at the same time needing to represent the interests of Nvidia and its shareholders. China is a massive market. And, you know, do you. Do you imagine that everybody just needs to move on from. From China and focus on servicing demand in the US and, and our allies, or how do you think it should be approached at this point?

Yeah, it's a tough situation for Nvidia. I think the line that they use is if you set the speed limit at 60 and we're going at 55, that should be fine. Right. Like, we're just trying to stick by the parameters that you set in these export controls and sell chips that are compliant. And then you have the government sort of making these big changes to sort of after Nvidia has designed these chips that is like, ready to start selling to the market, making these big changes that prevent their market can access overall. So, you know, if I'm a big Nvidia shareholder, I'm like, pretty annoyed about this. Right. Okay, let's. Let's have like, better proactive policymaking around this stuff. Like, I'm pretty sympathetic with that perspective, but yeah, I think realistically, you know, Nvidia is a publicly traded company that is access to the revenue. They want to sell chips into China. They're not sort of seeing the same national security concerns that a lot of people in Washington are seeing. And to be frank, a lot of people in the AI industry directly as well. Whether. Yeah, I think there's some question marks around whether they have been fully responsible with sort of doing their due diligence for shipments that are going into China and ensuring that like, control chips aren't being smuggled and that they're sort of complying with the letter of the law.

Yeah.

Overall I think we just need clearer policymaking ahead of time so it's easier for them to make these decisions.

Do you have a sense of what the black market market for GPUs looks like? I imagine, you know, you guys made a recommendation to add sort of like geolocation to these high performance chips. And I'm curious if you have any insight into how chips are flowing around the world in a way that, that may not be aligned to your.

We've been checking this for the past couple of years. So we put out a report in late 2023. They sort of looked at this question of okay, we've got these chip export controls now. What sort of scale of smuggling should we expect? And at the time, based on sort of news reports mainly the sort of cases that were being surfaced were just sort of single digit number of GPUs at the time. So at most sort of like, you know, a single server, like eight GPUs being smuggled at once. And now sort of like a bunch of reports came out last year that this is now just a huge industry. Some estimates is there's sort of around 100,000 GPUs went into China last year, controlled GPUs. Some people think it's much more. There's individual cases that have been uncovered where there's over 20,000 GPUs being smuggled at once. So this is some pretty serious shipment sizes like going into the hundreds of millions to billions of dollars. And there's likely a lot we just don't know, especially for these large orders which is sort of happening through professional operations behind the scenes. And in as much as it is happening, this shouldn't be a surprise. You know, China has a long history of smuggling controlled goods despite sanctions and export controls. And you know, it's proven that they were instrumental in helping to smuggle chips to Russia over the last two years to support the war in Ukraine despite sort of U.S. sanctions. So this would be like pretty on brand with the China playbook for this kind of thing.

How far behind do you think that China's domestic chips really are? Right. You know, we just did an entire backstory on, on SMIC and Huawei and things like that for, for our audience. But. And we had Aaron get on the show yesterday talking about how the gap, the gap in capabilities is maybe not as significant as companies like Nvidia or TSMC would, would have people believe. But I'd love to get a sense from you of how, how wide the gap, the capability gap is.

Yeah. So I think there's two dimensions to this. First is the quality of the chips themselves. So if we look at that, you know how, how many flops you have per second, the bad earnings chips that are coming out now are sort of three to four years behind the best stuff that's available from Nvidia. But you know, that has been a move towards catching up over time. I think export controls are the one thing that could reverse that trend as these have only really just become effective over the last year or so. The other, and this is sort of a really important factor is not just how good each chip is but how many they're able to produce. So you know, in AI especially it's not so much a matter of quality of chip that that is important. It's also very much quantity like how many you're able to produce and there it's where they're really constrained. So being able to produce high end chips by the sort of millions like TSMC is able to do is a very serious manufacturing operation. And China doesn't have that level of scalability because we have export controls across the stack with tooling and fabs as well. That is one of the key bottlenecks.

What do you think Chinese state investment into semiconductors looks like today? It's Obviously they've had 14 separate five year plans I believe around, around the industry. So at this, you know, at this point, I don't know how many hundreds of billions or you know, who knows, even dollars have gone into the industry. But I imagine this is, you know, one of the, you know, will just continue to be more and more important every single year.

Yeah. So the way this has been described.

Hello.

To revitalize the American domestic semiconductor manufacturing industry over time. China is doing this, this sort of equivalent amount of funding roughly every year to support its own domestic industry.

Tim, we, we keep, we keep losing you by the way. You're kind of cutting in and out. Maybe there, maybe the cyber attack has shifted onto your personal mobile device. They really don't, they really don't want Tim to talking right now because he's dropping bombs. But could you repeat, I think we have you now. Could you repeat what you said your potential estimate was?

Chip them in the States. This big landmark piece of legislation to revitalize the semiconductor industry here, that was about 52 billion as sort of a once off package that's being distributed over a few years. Best sources I've had say China is distributing like this equivalent amount of money roughly every year to their own semiconductor industry.

How do you think Huawei is, you know, reacting to the tariffs? They're a private company, so I feel like we have a lot less just information on, on, on a lot of things. But are they, you know, approaching some of these other companies, you know, nations that are facing tariffs specifically with kind of like new op, you know, basically trying to take advantage of the situation?

Yeah, I'd be doing in this situation. So similar to how Huawei managed to outmaneuver The United States with 5G, installing base stations all over the world and sort of surpassing the US in that technology, a similar play that they could be executing now is trying to do the same thing on data centers, especially if US equivalents are becoming much more expensive. But now they don't have comparable technologies on the AI chip side, but they certainly do on the lower end stuff. Alibaba Cloud, like Huawei, they all have their own data center offerings that they're sort of pretty aggressively courting the global market with.

Makes a ton of sense. What else are you thinking about right now in terms of now that the news came out that the order is being stalled, what do you think the admin should be focused on over the next weeks and months?

Yeah, so I think that one big lesson from all of this is that we keep making the same mistakes in the sense we keep acting too late to really be effective. The thing with export controls is you want to act early because there's always the risk of things like stockpiling, like if there's a room that Xbox controls are coming and they don't happen for another six months, it just gives a huge opportunity for companies to take preventative action and you can completely ruin the effectiveness of what you're trying to do. This has already happened with HBM memory. So like high end memory, that's really crucial in AI chips. The rumor was leaked that controls for this stuff was coming, and Huawei reportedly stockpiled around 2 million chips worth of this before the new rules came into place. We've also been caught, blindsided, you know, by Deep Seq finding out that they trained their models with US chips that were sold before the restrictions and policymakers being like, wait, what's going on here? The sort of SMIC, when they came up with their 7 nanometer breakthrough in 2023, it turns out they were using secondhand DUV machines, which we have since controlled, but they already got their hands on them. There's been a bunch of cases like this and really? Yeah, we've just been much, way more reactive than we need to be. I think if you just had a team within government who was technically competent and actually see and what's really important, going to be really important for AI technology in six months, you could actually be doing much better productive policymaking. So one thing we're advocating for is just setting up this kind of situational, situational awareness capacity within government so we can get it right next time.

Yeah, that makes a ton of sense. Well, we should have you back on.

When we have video. The Internet is not under attack. Apparently Google Meet is down as well. I don't know exactly what's going on, but story still developing. But I mean, thank you so much for taking the time. We really, I really do want to have you come back for a much deeper dive into all of this because it's, it's fascinating.

Yeah, let's get it on the calendar today.

Absolutely.

Do this properly. But thank you for making it work, Tim. And thank you for, you know, following the situation so closely and trying to get us on the right path here.

No worries, guys. Catch you later.

Cheers.

Talk to you later. Bye. Awesome. This is great.

Everything is down.

Everything is down. Down Detector is showing that not only is Zoom down, but Google Meet is down as well. I thought we could shift over, but we will have to figure it out. Let me message the rest of our guests. Zoom down. BTW, Google Hangouts 2 working on a fix. Pretty, pretty crazy time. I don't know if there's anyone actually reporting on what's going on. All I'm seeing is that apparently.

Well, I have some reporting. OpenAI is buying talks to buy Windows.

Yeah, I was about to say we should just talk about that. Maybe we just do the timeline and we get the guests to come back tomorrow and just do a stacked meeting because it seems like we are absolutely under attack.

Yeah.

And hopefully, hopefully it resolves tomorrow. But we had two interesting companies coming on. Captions is launching some stuff today. That's the app that actually Ev Randall over at Kleiner invested in Absolute Dog and I use that app all the time. Anytime you see me post a video that has some captions on the bottom. Always captions that. And then Dec announced a big fundraise today, something like $12 million. And we were excited to talk to you but instead we might just talk about the timeline because. Talk about the timelines working. So yes, this is a Scoop from Bloomberg. OpenAI is in talks to buy coding app Windsurf, formerly known as codeium for around $3 billion. Per source is familiar. It would be OpenAI's largest acquisition to date. They haven't been super acquisitive, but it makes sense. It's like if they need a wrapper in every single space, maybe they should just go and acquire them. We've been hearing rumblings about like a code first model, something that's really, really trained. It seems like OpenAI on the API side has become at least this is the vibe on Axe is like, oh, if you're going to use Cursor, you're going to use Claude. Right. And for most consumers there's still, yeah, ChatGPT app. I use 4o or whatever. But for the developers who really care about the quality of the code, that little extra push is what's valuable. And so having both a model that's trained by OpenAI to be specifically good at programming and then also having something that's popular, like the wrapper built on top of it, all the integrations that you get from Windsurface makes a ton of sense. So Zach DeWitt over at Wing VC, I believe. Yeah, partner at Wing VC. I clocked it. Good memory. He says incredible outcome. In 4 years they did a seed 3 million on 26 post by green Oaks. Green Oaks.

Another billion dollars for Green Oaks.

Another billion dollars for Green Oaks.

Yes. Quick plug for Patrick o' Shaughness latest episode with Neil Mehta over at Green Oaks. You can go find the episode Finding Future S&P 500 companies on invest like the best. Go check it out.

Yeah. Then they did a series a 25 million on 215 post again by Green Oaks, doubling down. Then Kleiner comes in for the series B65 on 510 million post and then $150 million series C at 1.25 post by general Catalyst. So founders and the team likely own 55%. So that's over a billion dollars in liquidity to the founders and. And the team. And at 20% of 3 billion, Green Oaks will be just shy of that billion. So.

Yeah.

So, you know, a small size gong for.

I would assume they own, do you think?

Oh, maybe they did like pro rata.

No, but I just assume they. They bought, you know, they continue to buy up.

So I don't know. I mean, 0.25 of 30. Was it 3 billion? Right. It's 750 million. So something around there. Not bad. Not bad for. For a $28 million investment.

Yeah.

To go from 28 million in to 750.

Yeah. It's interesting because you often see Green Oaks doing some of these later stage rounds.

Yep. They're known for like growth and being very analytical. But. But doing this one from seed, they got into seed. And the series A you love.

Very impressive.

Your jacket is going viral right now. People love it.

Yeah, we should just. Since our good lighting is still up and all the other platforms.

Yeah. Just do some. Let's just do more ad reads for the rest of the hour. Yeah, the back is great. Lucy and ramp, we put, we put our sponsors on here. We also just put companies we're affiliated with.

We still have, we still have space. This is gonna be.

That's legacy. So that's. Yeah, that. So, Yeah. I mean, if you're in the audience and you got some dough, maybe you want to get on the jacket. Now's the time to email Jordy. Get.

Get pretty much out of inventory.

Get out your ramp card and sponsor us. We are 100% corporate backed, by the way. Just so you know. You never ask our audience for a dollar. We're not audience captured. We do what's in the interest of the corporations that sponsor us. Independent media is dead. Corporate media, as I call it, is the future.

Free media for all.

Anyway, speaking of corporate media, the media around Blue Origin. We should go back to this because we didn't get to finish the conspiracy theory.

Can we play the video?

Yeah, we can probably play the video if it's up. We'll have the. We'll have the team try and pull that up, but. So the door opens to the Blue origin capsule as it descends. Then they say, hey, don't open the door. Close it. And then finally they open it up and everyone comes out. And so this was seen as proof that the mission was faked. And there is a conspiracy theory here. But my conspiracy is a little bit more anodyne. It's that they wanted all the cameras to be turned on before they did the big reveal of like getting out of the spaceship. And so they open the door. Yeah, you can see they open the door and then they close the door and not yet because, you know, NBC is here, but CBS hasn't arrived yet with the truck because these capsules, they come down, you can't just be right there. They don't know exactly where they're going to land in the desert. So if you watch the zoomed out feed at the very beginning, I think the very beginning of this, you'll see all the trucks are driving. They're scrambling to get to the landing site. Bezos was clearly just the first one in the chase vehicle to be Like, I'm going to get there first, make sure my wife's okay, you know, check in on the girls. The girls. The girls chat as they said girls.

Group chat.

Yeah. He sent the group chat to space. Send the group chat to space. But El Bezos falling down the hole. But he gets right back up. He gets right back up. That is crazy.

Oh, my gosh. That is a big drop.

I hope he's okay.

Drop.

I hope he's okay.

But isn't that what Brian Johnson is doing? Deathly afraid of falling.

Oh, really? I didn't know that. Yeah, I thought he does leg press. I think it's because he doesn't do that much leg press.

So he's worried about Andrew Coat Cote. Cote says, can any space nerds help me understand how the Blue Origin capsule came back down without any re entry burns? Was this a trajectory thing or some kind of superior material? Used Falcon crew capsule for comparison. So Scott Manley jumps in. He says it only flew at Mach 3 fast enough to reach space for a moment.

And to be clear, Andrew is trolling. He knows what's going on here. He's baiting the timeline to make it be like, oh, it's a conspiracy. But he, he. Everyone's like, laughing like, this is like S tier bait. Because, like, he knows that when, like this mission, I mean, they've done these missions, like all the time, and they never get burnt because they don't go that fast. And so that's what Scott's enumerating here. He's, he's laying it out. Scott says it only flew at Mach 3 fast enough to reach space for a moment. Not fast enough to stay in space. For that, you need Mach 25. As you can imagine, there's a huge difference in the kinetic energy between those speeds. And so the real conspiracy theory here is that this was a, you know, a fun space tourism marketing for space tourism. It's marketing for Blue Origin. They did a great job of puzzling together what is the crew that would probably go the most viral. So it's not just Lauren Sanchez, because that's probably not newsworthy. You got to put Gayle King on there. So CBS is bought in, but then also throwing Katy Perry, because throwing Katy Perry, you create an even wider audience of, you know, she's a global celebrity. Everyone needs to know about Katy Perry going to space. And it just seems like it's a lot of fun. So I would definitely go on one of these. I think this seems like a really fun trip. It's only a quarter Million dollars.

So I was thinking they're basically given space trips away.

They are, they are. I was thinking like there might be a trend of like, send the girls chat, the group chat to do something extravagant like this. We might be seeing this with other tech people. Maybe you don't own a space company, but maybe you just.

Yeah, guys out there, if you want to have a boys weekend.

Yeah.

Send up a girl's wife and her friends to space.

To space. But then they're only up there for 12 minutes.

Yeah, 12 minutes. I mean, still needs to be like.

How about, you know, cross the ocean on a sailboat, then the boys can go do the Nurburgring. Get them a wanderer for a month. I don't know.

This was funny. Jeff Tang.

Oh, yeah. This is amazing.

So this, this Icon, you had the founder Ken Davidson on the phone before on the show. He is a funny guy. He put out a competitor landing page with a comparison page and it was pretty hilarious as. As every competitive sort of comparison page should have. Icon checks every single box, including 100% money back guarantee, script writing and creative briefs, audience research, video ads, static ads, creative storage, asset tagging and splitting lip sync, UGC videos, creative analytics, competitor ad spying, ad manager upload, tier one investors, tier one engineering team, seven day work week. And he did not give OpenAI credit for tier one investors or tier one.

Engineering team, which is brutal because it's the same investors.

Well, and they're probably. Icon's probably using various chatgpt.

Oh, for sure, for sure. And also like, yeah, it's like founders fund is in Open. Is in OpenAI and icon. And so it's just such bait to be like, yeah, we're the only one with tier one investors. Like no one else in the AI category has tier one investors.

That whole AI thing.

Like this is true in other categories. Like we talked about general matter. Like there are of a lot, lot of.

Let's pull up the website too, because I actually think it's Icon. I think the Icon me website is like fascinating in general. Pull it up real quick.

So the story here is that, is that obviously Jeff Tang is very much like trolling Ken and being like, this is a ridiculous comparison. I love comparison pages, competitor comparison pages, because this is hilarious. And then Kenan says, what a lame comparison chart. This seems to be the website. And so he's like dunking on his own thing just to link to his own website because he just wants the traffic. He's like a total like just drive the bottom line, doesn't matter. If it's embarrassing, just make the money. And Jeff Tang.

Okay, so look at this website.

Oh my God.

This is the most conversion optimized.

This is hilarious.

It's actually so. It's actually so chaotic that I. Investors, Tier one team, Tier one engineers. I've never even heard the term tier one engineers and. But look how optimized this is.

Wow. Us versus them. There it is. They updated it. They updated. They put OpenAI tier one investors. And wow, they must have gotten.

And it's interesting. They're selling the product as you can become a founding advisor. What to icon.

What it's like, it's like almost like MLM coded. It's like a pyramid scheme now.

I don't know. I don't know. But they also sell. They sell.

You can just sign up for this, right?

You can go and app like go on the. Go on the pricing page. Just click. I'll send it in the chat.

Yeah.

This is the most chaotic SaaS website I've ever seen.

It's so funny. This kind of wild.

Well, look at the pricing page on here. They basically built it like a. Basically built it like an E commerce.

Yeah, it looks like checkout on E commerce.

So you can limit it all. There's only four spots left. Ends for 18.

No, no, no. For a SaaS product.

And scroll down. They sell a premium banana.

Wow.

I guess a banana you can just check out with Apple pay.

This is so hilarious. I mean at this point it seems like with him, with him like retweeting these and engaging these, like he's in on the joke. And so it's gotten way funnier. Like if this was just. Oh, he's really, really serious about all this stuff. Like he really thinks that he has two OpenAI.

Doesn't they got Saquon Barkley in here too?

Did they? Who knows? That's the thing with like this comedy landing page. I don't know what's real anymore.

Hopefully the product is real.

Hopefully the product is real. I mean we did talk to the Ridge guys. They said they were using it. It was pretty useful. I think. Obviously it's an early stage company. Kennan is trying to get some eyeballs, get some attention and just kind of break through the incredibly crowded market of AI tools. Right. And so you gotta make some noise to be able to just get through. But we'll have to put that whole site in the truth zone, figure out if Saquon's really in icons. Icons.

So far to our knowledge, he's in Anduril and Ramp. That's what he's a two company portfolio.

Yeah, but he could be ripping checks all over the place, all over the Valley. We'll have to figure it out. But speaking of ramp, time is money save. Both easy use corporate cards, bill payments, accounting, and a whole lot more all in one place. Go to ramp.com, switch your business to ramp.com.

Switch your business to Ramp.

And Ryan Peterson had a good post. He said a 1% daily improvement compounds over a year to a 37x improvement over the entire year. But I think he's thinking too small. So I asked ChatGPT, what does a 100% daily improvement compound to over a year? If you compound by a hundred percent every day, you're doubling each day. That's exponential growth, baby.

That's right.

So if you start with one as a base and double it every day for a year, by the end, you will be 1.52 times 10 to the 109th power better than you were at the start of the year. That's 1. Followed by 109 zeros or 1.52. November in the short scale, doubling every year, every day for a year gives you 1.1.52 November decillion times your original value. This is absurdly huge. A classic example of why compounding growth explodes with even modest periods.

You don't understand.

So if you're thinking about improving 1% daily, just do 100% daily. And then you'll just be so much better off. So risk on it's great soundboard. They haven't cacked the soundboard.

Yeah, it's been our most chaotic show today.

It has. It has.

Facing cyber attacks.

Yes.

We have a new soundboard.

Yep.

Zoom is down. Google hangout is down.

You're not in a suit.

I'm not in a suit.

I think it's a more relaxed show. We're just having fun with it. It's okay. It's okay. The show must go on. That's the most important thing.

Well, you know what? Service is not down.

John, get bezel.com.

That'S right.

Go to Bezel Concierge. They're working overnight. They're working overtime. Your Bezel concierge is available now to source any watch on the planet. Seriously, any watch. Pick up a. You know, they're not gonna. They're. They're not gonna hack your GMT master, Batman. They're not gonna. They're not gonna hack your Submariner.

Yeah.

They're not gonna hack your Paul Newman daytime.

I'd like to see him try.

I'd like to see him try. It's gonna keep your, your Patek Philippe cubitus is still gonna tell time after the AI overlords have destroyed all of the data center infrastructure in this country. But we'll see.

Should we pull up this post from Josh Kushner?

Yeah, sure.

He says, so we talked about this yesterday.

Yes, yes, yes.

Somebody asked, do VC funds invest in competitive companies? Kyle Harrison pulled up a chart that said evidence would point to yes, showing how many of the big venture capital firms have invested in multiple or at least a couple foundation model companies. And Kushner over at Thrive Capital has. I don't even think he's on the chart because he's only invested one and he's invested in a very important one. And he likes, I think, staying off of the charts like this for the most part. I think benchmark or, sorry, Jeff Lewis and the team have also just gone all in on OP AI. So there's still quite a few. But again, I think that this is an evolution of venture capital and it is a natural evolution that firms will invest in competitive companies. And I do think that firms generally do a great job of keeping firewalls up and not sharing information that would be damaging to any the individual companies.

Yeah, it's interesting. I mean, I think that obviously there are secrets in these foundation labs and there are algorithms and just, even just the path down the R and D tech tree, like reasoning models came from Ilya Sutskever working on q star@OpenAI. When he leaves to start SSI, you know that there's going to be a reasoning component, a reinforcement learning component on top of LLMs. There's going to be self play like this worked well at OpenAI. He takes that and you can't patent that idea of like a reasoning model or applying reinforcement learning on top of LLMs. And so not only can you not control that or patent that or keep that out of the hands of GROK or Thinking Machines or Mistral or any of these other companies. But I don't think that the VCs are the vector for these types of leaks. I think it's like the AI happy hours and the parties in San Francisco where people are talking and they're like, oh, did you see this paper that someone just published? Like, yeah, or somebody, you know, reimplemented the open source.

And I think this is founders in the early part of their career. I, I've seen them be worried to pitch a firm because they were invested in something that was competitive or adjacent to what they're doing. And the reality is, is your competitors are going to find out about what you're doing. Generally.

Yes. And that's not necessarily your moat. It's rarely like the Coca Cola formula rarely applies. But at the same time, like, it is reasonable to say if I'm, if I'm really partnering with a firm and they're going to take 2030, they're going to build a big position, they're going to be on my board, do I really want them to have this weird, like, incentive. Yeah. This weird conflict where they find a great VP of engineering and they're like, well, I could send it to you or I could send it to your direct competitor that I'm also invested in. Like, any sort of like value add does get, does kind of get spread thin. But I guess it, I guess it gets back to this idea of like, how commoditized is the foundation model layer and how different will these companies look over, over time? I mean, you know, Andreessen here is in OpenAI and XAI, those two companies seem at each other's throats like, like OpenAI is launching a X clone. XAI bought X and so they're like really direct competitors. But at the same time, like, if OpenAI is buying windsurf and developing more consumer products and then SSI is saying, hey, we're staying out of that entirely. And then Mistral seems to be focused more on B2B and government implementation and Anthropic is very much on like the API side, for example. You could imagine that foundation model just becomes like cloud. And at that point, like, you wouldn't be that upset about investing in like a Salesforce and a PagerDuty and a Twilio. Like these are very different companies over time versus if they be cut clear.

The end state is not just 10 yes, you know, chat apps.

Yeah. So it's like the foundation model layer may commoditize, but that doesn't mean that the companies are to going to commoditize. Like Lyft and Uber are true, true supplements to each other. Like they are, they are not complimentary. And in any way, you're either spending money on Lyft or Uber. They're very similar experiences now, one way.

Better than the other businesses.

For a long time it was a.

Capital fight because people would open both apps.

Exactly, yeah, it was a capital fight.

Which is faster, which is cheaper.

Yep, yep, yep. But, but if, if OpenAI winds up looking completely different than thinking machines or SSI or Mistral. In a few years that like, there could be winning companies that are very different but just built on the same, you know, foundational technology. Yeah.

And I, and I assume that we'll see even some of the foundation model companies merge. Yeah, I mean, we're yet to see that because they're still extremely well funded.

Yeah. Didn't. Was it Sequoia that did intel and Nvidia? I mean, they're like a decade apart, but in theory they're both like chip companies. Right. But they operate in very different spaces and they have very different models now. Fabless versus integrated. But, you know, there are some funds that have stayed the course. Josh Kushner over at Thrive is serial monogamist. Delian also chimes in to say, sorry, can someone explain to me why we're on this if we only have one arrow?

Yeah.

Doesn't that mean we're not spraying and.

Praying, manage to stay off the list? Founders fund was not as clocked on there.

Yeah, but it is, it is weird that like this is. I mean, I guess it's just, I guess this original, this original diagram was just to show like what funds are in what are in what foundation model companies, but then very quickly became, you know, oh, they're investing in multiple. But. And then it makes less sense for FF to be on there. Anyway, did you see Black Flag launched today? The founder sent me this. It's an incubator for hard tech companies and they are split across a variety of different industries.

Well, John, I'm actually launching an incubator for incubators.

Oh yeah.

So I'm going to reach out to these guys, see if they want to be a part of my incubator.

Incubator, the meta incubator. I mean, I think that's just being called the lp. But Harpoon Ventures is the, is the main firm behind this and they're in collaboration with SHIELD Cap vc, who we've talked to before, and in Q Tel is an investor, which is the CIA's venture fund. So that's pretty, pretty fun and good news. I mean, I think that there's a lot of, a lot of folks that, you know, I think you could say, like YC is the incubator for defense tech and YC has been open to this, but at the same time, like, there might be some benefits to having an incubator for like, specific hard tech companies. So, you know, you maybe amortize the cost of some like CNC machine or some like, you know, shared hard workspace or some capex stuff, but I Don't know if they're planning to do that.

No, I think it's awesome. I just want to see just more.

Of this stuff is great. More competition. You know, companies are now going through multiple programs. I mean we talked to Ali and it seemed like some of those companies had done, had done NEO and then also done Teal Fellowships and then Teal Fellowship Fellows and then they also raise yc. Exactly. It's like, yeah, there's an overlap of these. There's just like at the end of the day it's just like for, for Black Flag.

I don't know anything about it other than this graphic. But what's cool about specifically defense and aerospace and anything government related is that there's an entire. They can provide a very clear benefit to founders that maybe don't have a defense aerospace background.

Yeah, yeah.

But can sort of. They can really accelerate that knowledge and networks and things like that and allow people that maybe we're building in SAS to come in and play in a new arena. So.

Yeah, well, if you wind up with 20k from one of these incubators and it's no strings attached money, why not throw it in public?

That's only neo by the way. NEO is the one that, I mean.

We'Ve heard stories about Teal Fellows investing their, their money if they're profitable. Because it's, it's, it's kind of no.

Strings attached to recommending venture dollars. These are grant basically grants.

But I mean why don't we have an accelerator for hedge funds? Like you go in, they stake you. I know they exist but like you know, you just get the first like high frequency trading shop from yc that's going to be a banger company. Come in. It's like your demo day slide is just like we started with 500k from YC AUM's 5 million. Now we 10x that.

There we go.

We got a couple more 10xs. It's a billion dollar company. Get in now.

There we go.

Here's the LP agreement. We're not doing safes.

No safes here.

That'd be funny. Anyway, you know about public.com investing for those who take it seriously. They got multi asset investing industry leading yields and they're trusted by millions, folks. Anyway, I think we've been through most of the timeline. Most of the. Great.

This was great. This was a chaotic show. I'm excited to figure out why all of our, you know, Internet services.

We need a polymarket on this right now. We need a polymarket. Why did Zoom And Google hangouts meet video, whatever they're calling it these days. Why did all the, why did all the services go down? Was it a cyber attack? Was it not? I want to know about it on Polymarket. That's our other sponsor. So go to Polymarket and check it out. We'll try and get a market up asap.

And I just want to say I've, I think I've gotten at least 30 individual messages about the TVPN jacket from. People really are wanting it.

That's amazing.

We weren't even planning to release, to.

Sell these, but should we, should we tell them it's. It's 9.99. But financing it will be available. You can buy now, pay later. It's. It's 12 payments of $99 if you want one.

One.

So that's just kind of what the market will bear.

Yeah. And we're going to donate the profits to all in, which doesn't run.

No, no, no. We're going to donate the profits to my next watch on Bezel.

That's right. Okay.

Get bezel.com, folks. They got concierges. They'll find you a great watch.

John loves watches so much, he's running two bezel ads on the same show. But thank you for tuning in, folks. I'm excited for tomorrow's show. Yeah, we want to talk about who's coming on.

We're coming back. We're going to bring on tons of people. We're going to stack it. Maybe we'll do a six hour stream to make up for the lost half hour. Like, if you lose a half hour, you owe the fans four hours.

That's right.

Right.

Every half hour, it's four hours.

No, I mean we, we got a great show tomorrow. We got Delian coming on. We got Sham Sankar from Palantir.

CTO of Palantir.

CTO of palantir. First Ford deployed engineer. He's been the company for like 20 years. It's incredible. I, I was hanging out with him, it was public company. And I was like, so what's next? Like, you're a public company. Like, what's your goal? And he was like, oh, we got to get into the Fortune 500. And I was like, that's an insane goal. I've never heard anyone like, frame it like that. And then like a couple months later, I had to send him a congrats message because it was like, they are now officially in the Fortune 500. That's sick. Congrats. And then we're also gonna have some robotics CEOs come on, work on a couple different robotics companies and talk about those and of course we'll, we'll be bringing you to the news. Talk about whatever's top story. I think we covered the Nvidia stuff pretty well but I'm sure there'll be more development.

I just realized we, we completely missed a major story.

What's the major story?

Figma.

Oh yeah, the S1 dropped.

Did an S1.

Congratulations to everyone who's been working on Figma.

Amazing milestones.

I mean it is one of those rare companies where it is an overnight success. It's been like what, 14, 15 years but overnight success. But congratulations to Dylan Field and all the folks over at Figma. What a roller coaster ride. Years in stealth back and forth with Adobe.

Wasn't it like four years?

Basically they shipped a product or something like that and they were still raising money. It was sheer force of will. Classic overnight success. You love to see it. And we're gonna be diving in deeper to Figma. We should tell the whole story. We should do the whole. A whole deep dive on the company. This is a fascinating story.

And we'll be at config.

Yes we will. We're gonna be doing post game interviews from Configuration, interviewing some of the greatest designers in the world. Some entrepreneurs, some investors, all sorts of folks in the design world. That should be fun. I'm newer to the design world than you. Not a. I'm not a Figma daily driver like you, but I'm excited to learn a lot about it.

You're a creative athlete though.

Yeah. Creative athlete, Yeah, a little bit. I mean there's creative stuff all over. Design is in everything in this world. Anyway, thank you for watching.

All right folks, we will see you tomorrow afternoon. We are going to call up Tim Cook.

Yep.

David, Daniel Elk from Spotify. Try to get these services back online.

And Sundar, because Google went down. That's unacceptable. I think our SLA says no downtime.

And who's the Zoom CEO again? It's Eric. Eric, you on?

I just know him by Erica. When I text him, I just.

Yeah, we'll text.

What the hell is going on?

Put them all in a group chat and we'll say let's get it together guys.

Let's get it together. Let's step it up. Now is the time. You can't be slacking off. Yeah, can't be slacking off, guys.

Every minute of productivity counts.

It does.

All right, have a great afternoon. We'll see you guys tomorrow.

Bye.