AI Boom or Bubble? Plus the US Government Reopens After Record Shutdown

Hello and welcome back to the Market Maker podcast where I'm joined by Piers Curran, who's currently sat in a booth in New York City. So on the other side of the world at the moment for me in London. But we've got 3 things we want to talk about on the show this week. And number 1, Trump has signed the deal ending the longest government shutdown in US history. But are we just kicking the can down the road toward another fiscal cliff, i.e., the end of January? We'll look to explain. Number 2, oil had a pretty big dive. I think it was around 4% on Tuesday of, or Wednesday of this week. And it's very much a supply side story, but Saudi's Mohammed bin Salman, MBS, is meeting with Trump and a $600 billion geopolitical power play could be in the offing. And how might that impact production rates and subsequently oil prices? And then finally, are we in an AI bubble? It's the trillion-dollar question, not the million-dollar question. Goldman Sachs does not think so, at least not for now. They have come out, I think it was their chief equity strategist was being quite bullish this week. So what is the rationale there? So they're the three things. So Piers, how are you?

Not bad. Bit jet-lagged, I have to say, but you know, I'm in a booth. I'm in a WeWork booth. I don't know if listeners have ever used WeWorks, but you could be literally anywhere on the planet. The WeWork phone booth is exactly the same. You know, it's like you're, I don't know, there's some kind of comfort for some people like McDonald's. You know, you go and get a Big Mac, you could be in Tokyo or Johannesburg or Chicago and it's going to be the same. And so it's kind of a bit like that with WeWork, this cookie cutter thing. Yeah. So I am in New York, but right now I'm sat in a booth that really could be anywhere on the planet, to be honest.

You're in the TARDIS at the moment. So, well, look, let's talk about this government shutdown. We have talked about this before in recent episodes, and the last time we were talking about it, we were talking about it in the context of the markets didn't seemingly didn't appear to be too worried. And at that point we were already pretty convinced that this was going to be a record-breaking length. So let me just give you the market reaction and then we can go into some of the finer detail. So brief rally, fairly muted though, with equities markets were pricing in resolution as we were talking about before. Rates, short-end yields, bit jittery. We've still got a lot of economic data delayed still to come out to define the Fed's kind of rate path going forward. Volatility, the VIX went down, but longer-term vol expectations crept up. So what does that mean? There's a couple of words there. So obviously just to kind of demystify some of this terminology. So short-end rates, VIX going down, long-term vol expectations going up.

Yeah. I mean, Well, look, firstly, I think, was it last week on the podcast we said we're just hearing nothing about this shutdown? And so then our conclusion was this is definitely going to carry on. It will be the record longest. There's no noises at all. What I would say is what has happened is it's been a very, very quick turnaround from hearing nothing to actually Trump literally signing on the dotted line and the shutdown ending for now. And so I think there was a little bit of a relief rally when you look at equities, right? So equities did go up. I think there was more just because all of a sudden it came super fast. You know, it's not the case that markets were pricing for a different outcome. They were pricing for this to be sorted. But just not perhaps quite so soon. So there was probably another few weeks more worth of sort of anxiety in the markets. But look, this is such a boring thing, this US debt situation and the government shutdown. It's just rinse and repeat most years. Yes, when Trump's in office, the political situation's more polarized, and I think therefore then it's harder for these two teams to agree on terms and move forward. And so I think with Trump in the seat, it's a bigger issue than perhaps normal other presidents. And look, this is, as you say, this is kicking it down the road. So January 30th is now the new deadline, and you can bet probably good money that the government's going to shut down again. So yeah, equities went up a touch, short-end yields a bit jittery. I mean, I haven't read that article, but if that was a headline, That probably tells me the journalist didn't really know what was going on. Jittery. What does that mean? Is it going up? Is it going down a little bit? I mean, the thing there to think about short-term yields and why maybe they're focusing on that is because ultimately what's the economic impact of this government shutdown? And there is one. And when there's talk at the moment, estimates are that it'll strip 1.5% off Q4 GDP growth. Yeah, that's not a small number. Right? So therefore, if the economy's going to suffer, well then what might happen to interest rates? So this is where the Fed, and remember they got a meeting in December, we'll be talking a lot about the fact that one of the kind of repercussions of the government shutdown is you don't get any economic data. So the Fed are a bit blind. And we were talking about, or Powell was talking about driving in the fog. You know, what do you do? You slow down, right? He was implying he's really lowered rate cut expectations for December. There is still, I think it's, I think we're about 65% at the moment chances of a cut. So it's pretty close. Normally it's quite clear cut, right? So this is 65%.

Just had a look, 51 now.

Oh, it's changed, right?

Wow. 50/50.

Okay. Now that is interesting that you very rarely, I mean, look, we've still got a few weeks left until the meeting, but you very, very rarely going to a meeting anywhere near a 50/50 split, meaning it could go either way, cut or no cut. So if, you know, the longer that stays at 50/50, I think the more jittery markets will get, right? But my point about the short-term yields, if you're thinking government continues to stay shut, economic impact increases, the negative impact increases, more chances of a cut, you could say, short-term yields go down to price that increased chance of a cut. You might have safe haven buying of bonds as well. That drives up price and drives down yield. So that's that. I think with the VIX, so you talked about the volatility, short-term volatility dropping because, all right, well, they've sorted it out. Well, that makes sense. But then the long-term vol expectations creeping up. So that's about, yes, they've sorted it out, but They haven't.

So that's the can kicking. That's the technical reaction of can kicking.

That's right. Exactly. Yeah. Right. So, you know, we move on. The thing is, it's such a, the reason why it annoys me, it's such a political, it's just political gamesmanship, right? And look, there's the midterm elections in 2026 and in the Democrats' mind, it's like, can we Can we push this problem and risk, can we get it as close to the midterms as we possibly can? Therefore it reflecting more negatively on the incumbent government because your average Joe on the street, oh, well the government's shut. Well, hang on, that's annoying. Well, who's the government? Well, it's the Republicans. Oh, it must be their fault. So the Democrats will be trying to just push this for as long as they possibly can. Purely for political gain for themselves, completely ignoring the fact that the longer it goes on, the more havoc it wreaks for your everyday person out on the street. They don't care. That's my big bugbear about it.

And this makes sense then of the language Donald Trump said before signing the bills, so they're all in the Oval Office and he said, the Democrats tried to extort our country, was the language that he used, which would be absolutely is what you'd expect given exactly what you said. Yeah, there was, there was two parts to this, I guess, more market orientated that I wanted you to break down, which is one sector impact and some of the fallout on the corporate side because some of the things I've been reading about are kind of twofold. One about airlines and travel, and we've obviously got Thanksgiving coming in just a few weeks' time, which is a big moment for the US economy from a retail and consumer perspective. And then on that consumer side, Snap, which is the Supplemental Nutrition Assistance Program, or food stamps. But why are we talking food stamps when it comes to thinking about the US economy?

Well, this is interesting, and I was walking along the street in New York yesterday and like on one of the buildings they had this kind of, you know, those old like ticker tape sort of LED displays where you have news headlines kind of scrolling along. They had one about this and I just stopped and stood and just actually read it that there are 40 million American people on food stamps that, that, so obviously this is the bottom end of the income bracket. And so they are given subsidies on food and this comes via these stamps. There's 40 million, right? That's a huge amount of people. So how, how does it happen? So this government shutdown has meant this SNAP, so Supplemental Nutrition Assistance Program, has, has closed because there's no one to administer it because they've gone home, they're not being paid, right? The government officials, that is. And so the lower income, they're not getting their food stamps. So what is it? Obviously they have to eat, so they've got to stump up the full amount of cash to buy the food themselves, which means they don't have— they then have less disposable income to consume and spend on other stuff, right? So, you know, so from an economic point of view, it It doesn't really impact companies that sell food to these guys because they're either buying them through stamps, which is the government paying, or they're paying themselves. So it doesn't really alter their revenues. It's just going to have that secondary order impact of retailers and particularly discount retailers, right? Will probably be, you know, that's the place where most of the spending goes from the lower income bracket. And when you're going into Thanksgiving holiday season, where you're buying gifts and all the rest of it, it's then, well, okay, they're spending less money in those shops. So it can definitely have a negative impact. I think that's why all of a sudden, I do think Thanksgiving just ahead of us, like in 2 weeks' time, I reckon that was the catalyst for all of a sudden this just got done and solved, or at least pushed down the road. I think it's, look, let's not shoot our own goal here. Let's just get back to work for Thanksgiving and then we can carry on arguing in the new year.

Yeah. And then the other corner of markets that is always being affected within these discussions is healthcare, health insurers, consumer discretionary spending that we've just mentioned. But yeah, on that health side, the Affordable Care Act or Obamacare, and this is the big contentious point as it always is and still remains unresolved and will rear its head again in several weeks' time. This is about, you know, for those not fully aware, that subsidized health insurance to Americans who don't get coverage through work and the shutdown core of the political fight, whether to extend those subsidies or not. So yeah, it seems like there's a couple of little bits here over away from the top line headlines that are quite interesting to watch as little tiny areas of sectors here from an equity perspective.

Yeah. And you mentioned airlines as well. And again, thinking about Thanksgiving, right? Everyone travels home. So it's like one of the biggest weekends, it's the biggest weekend of the year for travel companies and airlines, right? And the fear was, 'cause air traffic controllers, well, they're government officials, right? And so they, that's been really disruptive. However, 'cause obviously I've just flown to New York, right? And actually, I flew here on Monday, so the government shutdown thing hadn't been resolved at that point. And I was reading in the press, oh, this thing about Delta Airlines issuing a warning, their earnings are going to get dented. You know, Thanksgiving holiday's coming, it's going to be a disaster for us. I was reading stories about, you know, massive queues, you know, immigration and I was thinking, oh God, this is literally the worst possible time I could travel to New York. Do you want to know what happened? I turned up, landed, zero queues. I literally queued for 5 minutes at immigration and I was through and out the door and in a taxi.

Hang about, hang about. What airport in New York did you land at?

JFK.

JFK. You went through in 5 minutes.

I've never been through JFK quicker. So it's interesting because you latch onto that narrative. It makes sense, right? Oh, air traffic controllers aren't at work. Ooh, that's That means flights getting canceled. That means disruption at airports. That means Delta Airlines, et cetera, revenues dropping. It makes complete sense. But when you turn up and actually have a look, it's like, what's— where's the problem?

Okay. So who needs alternative datasets when I've got Piers Curran to send all around the world to—

That's right.

Boots on the ground.

Cool.

Oh, that's interesting.

You might have to square that one off with my wife.

But well, she went with you to New York, that might be a different story. But, well, sure. All right, so let's move to the second topic and that is one of crude oil. So Brent fell below $63, WTI near $58, a 4% drop that we've seen midweek. The core driver here, what has been a long anticipated supply glut, but I guess reality just hitting home somewhat. The IEA sees a 4 million barrel a day surplus coming for next year, and a lot of people looking at the forecast now for the year ahead. OPEC flipped its Q3 outlook from deficit to surplus. So that happened at the same time. And maybe just a quick refresher, if you may, which is the time spread WTI flipped into contango. So when we talk about time spreads for oil products, and then it, the shape of that moving to reflect contango. Just a quick, what is that?

Yeah, I mean, I think, I mean, if you look at a chart of oil, it's quite interesting because basically, obviously if you go back, right, post-COVID, we then have Russia invaded Ukraine and this triggered, you know, this really big sort of supply risk. Which shot prices higher. And you might remember in 2022, for much, you know, the big middle chunk of that year, the price of oil was over $100. And it was a big part of the inflation crisis post-COVID and all the rest of it, right? But really, ever since that 2022 peak, for 3+ years now, oil's actually been trending down. And just last week, or sorry, a few weeks back, we actually set a 4-year low for the price of oil looking at WTI crude. And you know, this news flow comes out, you know, about surplus supply in 2026 and it's like, well, okay, I mean this should really mean this downtrend set should continue and, you know, $50 kind of here we come kind of thing. So yeah, it is interesting and I think Ultimately, from an economic growth point of view, generally speaking, lower oil prices is good news for global growth. Clearly it's bad news for countries that are dependent on oil revenues, you know, the OPEC countries, for example, right? But for the net-net, you know, for the planet, it's good news if energy's cheaper. But it's interesting with the You know, I mean, contango. Yeah. As you say, this is basically meaning where expectations of rising demand or lower supply, it's about the future, right? And it's about storage. And so basically anything that makes future delivery more expensive, then you're expecting prices higher and this creates contango, right? So if the price of oil is more expensive in the future, but I still think we're in an interesting pattern of, because on the one hand you're saying, oh, increased supply, supply glut. But then we're going to talk about AI in a minute. I thought there was an energy crisis. I thought there is a lack of energy to be able to fulfill the compute power that this AI boom requires. So I can't quite square these two. How can we have an oil glut oversupply on the one hand, and yet some kind of energy crisis and there's not enough energy to fuel this AI boom? That obviously they're contradictory. So I don't know. I think that the IEA, well, maybe they've got it wrong. How about that? I mean, I think it's obviously very hard to predict 2026. We'll talk about AI boom in a minute. Is it sustainable? I mean, ultimately, I think if you want to bet on the price of oil for the next 6, 12 months, honestly, really, it's a bet on AI. And if you think AI is not in a boom, and if, sorry, if you think it's not in a bubble and this growth is sustainable and the energy requirements are true and there is a crisis, well, demand for energy is going to be outstripping supply, so the price should go back up. It's just, I think we're heading into year-end and maybe concerns about the AI boom sustainability are rising a little. And I think this is feeding in then to 2026 energy demand expectations, which is why you're getting the IEA saying what they're saying.

So yeah, I mean, from the broader perspective, Saudi Arabia, I mean, they have been increasing production or output with OPEC+. US production's been ramping up.

Yeah.

I mean, Saudi increasing oil supply, I guess from a top-level perspective, they've got their deliverables on their fiscal side for their revenue targets. I mean, I don't know if you caught the article in the FT about the line that they're building. I mean, that thing is just insane and doomed to fail to, I think, the vision that's been painted. I saw a chart of the line drawn on Britain, and I think it went from London to Norwich or off the coast. I mean, it—

haven't they shortened it though? Isn't it like going to be like way shorter than they were originally?

I think what it ends up being, I'm not sure, maybe like the Mexican border wall. But in terms of like the Saudi side, Obviously they're driven with lower prices, so that could mean higher volumes that could offset lower prices to maintain their fiscal targets. There's the defensive market share against the US who are also ramping at this point. But interestingly, Saudi Crown Prince MBS is meeting Trump next week on November 18th at the White House. I mean, that's red carpet treatment. Right there, because the next day on the 19th is going to be a $600 billion investment summit held in Washington, DC. And on the agenda, one of the top billing, of course, which is largely why their relationship exists in such a tight form, defense guarantees in what otherwise is a fairly volatile region of which Saudi sits obviously just over the Gulf of Iran. So yeah, I just thought this was quite interesting actually, and I just wondered your take on how does that translate that relationship then into oil and this Saudi strategy, Trump diplomacy kind of mindset?

No, I honestly think the Saudis, like behind closed doors, they're going, oh, this is amazing. This AI boom, it literally, because they Their 2030 vision, MBS, is about diversifying. You know, it's a long-term project, decade plus. They've been on this 2030 vision for more than 5 years, right? And so it's a long-term play to diversify the economy away from its dependence on oil revenues. I think, I don't know what the stat is, but it used to be 5 years ago that 87% of Saudi government revenue came from oil. So it's like, right, we've gotta get off this dependency given where the green agenda is going and how clean energy is really the future. So we're going to be in a bit of trouble, right? So they started to diversify. I actually think this AI thing means that actually, long-term demand for oil, it's gone back up in my opinion. And I wonder whether, and it hasn't really been said yet, but At the moment, the Saudis are like, yeah, we'll invest, we'll invest in the US, we'll promise $600 billion, we'll buy your kind of military equipment, all the rest of it. And for Trump, that's a great win and a great headline. I wonder if Trump is also being influenced by the big tech companies who are really worried about a lack of energy in the future and how the US, you know, the regulatory situation around building like nuclear plants. It's just a nightmare. And whilst you've got China building 100 nuclear plants in the next 5 years, the US, it like takes 10 years to get a permit signed off to build one, right? So these tech firms are like, this is a crisis. I wonder whether they're saying, look, you need to get the Saudis back in our sort of friendly terms here. And maybe we are going to need Saudi oil again, because at the moment they don't need it. They produce a huge amount of oil themselves and they're kind of no longer dependent on Middle Eastern oil, which has been interesting from a geopolitical point of view. But I wonder whether they're thinking that might actually swing back. I'm talking long-term here, right? Over the next sort of 5 years. So I wonder whether you start to hear noises about that from Trump.

Yeah, I'm pretty sure.

MBS meeting.

Pretty sure in that circularity ecosystem that there's some Saudi money awash that probably is not just on the Trump political side, but also, like you said, those tech firms. I'm sure there's a bit of capital flowing from that direction as well that they want to— it all feeds in, right? It's kind of everyone's dependent on each other, in this case, Trump and the tech bros, so to speak. Just fact check the stat. There's a couple of different sources, but it's looking more like the 50s% in terms of government revenues, oil and gas for Saudi.

Okay.

So that's perhaps has normalized a bit from what it was. But yeah, let's move on. Let's talk about final topic. And that is one that I think it came out last week as a headline, but the research report in its full entirety has come out. You can access that if you just go on my LinkedIn account. I did post about it with the link from Goldman's, but I just wanted to rattle off that title and some of the rationale, and I just wanted to agree or disagree kind of game with you. And if you disagree, for what reasons? So the title was, are we in an AI bubble? And Goldman Sachs doesn't think so, at least not yet. And so they talk about AI evaluations, spending, market structure, and so on. Gets distilled down to about 7 or 8 different points. Number 1 is one I think we've talked about probably already, which is about this is not a dot-com repeat. So why is it not a dot-com repeat?

It's not because the companies driving the bubble, as in the share prices that are going up the most, well, they're these Mag 7 giants who've got revenues and cash flows that are just off the scale. These lot are paying dividends, you know, even though they're spending tens and tens and tens and tens of billions of their cash on this kind of data center buildout, they've still got spare cash on top of that, right? So they are insanely profitable businesses with monster revenue growth. The dot-com bubble, that was valuations based off nothing. That was like valuations that went sky high based off future revenues and future profits. Yeah. There was nothing tangible in that moment. So that's why that bubble burst. There's a lot of tangible cash kind of parked, kind of propping this thing up. So that's the big difference I would say from the dot-com compared to now.

One of the things that they talked about in the report was about a gap between private and public market valuations. So why would people look at that gap and what's its importance and what is it showing at the moment?

Well, and I think this, so what's happened in recent years is the private markets have got a lot bigger. So that just means companies that don't go public, right? And like the big guns are like Stripe and SpaceX, for example. Where we'd been kind of waiting for their IPO and then for them to go public for years. They just haven't. Well, why haven't they? Well, why do they need to? Because ultimately, if you go public, you are then open to a lot of regulatory scrutiny. You've gotta do your quarterly earnings. It's just a pain. And so actually these days, a lot of companies are staying private. Now, what does that mean? It just means it's more opaque. You can't really see, you don't get earnings reports, so you actually don't really know what's happening there. And their valuations tend to be done off revenue growth because we don't get any detail about their profits, right? So it's more about their revenue growth multiples, whereas public markets, so that just means companies that are listed on the stock exchange, you know, they're Their valuations are based off their share price, which is a function of their profitability and their cash flows and all the rest of it, right? So if you get a divergence, and this is what Goldman's were pointing to, if you get a divergence between private AI firms and public AI firms, and if you get a divergence where the private ones, their valuations are growing faster, that's a sign of fragility. That's a sign of bubble. Bubbly type activity, right? Because you're getting the opaque stuff grow faster. It's the publicly listed companies, that's where's the transparency. So I think that they are right to point if that divergence increases, there is that you would say the bubble risk increases and you look at stocks like Palantir, which is up about a quadrillion percent in like one day, right? It's ridiculous. But so they are right to point to that as one line item that in favor of those that worry this is a bubble.

A lot of the people who worry about the bubble often talk about CapEx. I mean, pretty much we just had the tech earnings, right? And it's like Meta's CapEx and things like that is what people will jump all over. But I guess you could interpret that CapEx in two different ways, and Goldman have taken a bullish spin on this, right?

They have, can I just take a bearish one before? Well, let me tell you what Goldman, I mean, Goldman's are saying, look, data center spend will hit $3 to $4 trillion by 2030. These lot, they only started spending money, big money on this, you know, really in the last 2, 3 years, right? So 3 to, so it is the biggest ever corporate CapEx project in the history of mankind by many, many times. Okay. That, however, Goldmans are saying this is fine because the demand will be there. The return on investment will be there. They're saying AI could create a $20 trillion economic value. And $8 trillion of that they reckon could flow to US companies alone. That's their estimates. You get— Yeah, go on.

Quick stat. Big tech companies. So your Mag 7 types, they're expecting, I think this is more around this year, it's about $400 billion capital expenditure and data centers. So where we're at to give that $3 to $4 trillion a bit of context.

That's just this year though, that $300 to $400 billion, right? Yeah. Okay. I think what they're saying, $3 to $4 trillion by 2030, I think they mean the total CapEx spend on data centers adding up all the years. You know, I don't think they mean in the year of 2030 there'll be $3 to $4 trillion. But look, they're saying the demand will be there, right? My favorite stat, which kind of, you know, definitely falls into the camp of this is a bubble, is OpenAI's projected CapEx spend on data centers out over the next 5 years is projected to be $1.4 trillion. Their business has a revenue of $13 billion. $13 billion revenue company saying they're going to spend $1.4 trillion. That maths doesn't add up, right? So that's a great one. But then when you peel it back, yes, $1.4 trillion committed, but actually they've done all these circular deals with Nvidia.

Too big to fail, Piers. Too big to fail.

Well, indeed. So you could say that about half of that spend is actually going to be covered by their partnerships they've formed. So it won't be $1.4 trillion, maybe it's $700 billion, right? And then the point is there, well, yes, they're committing currently on the roadmap of growth that their expectations and their analysts are coming up with. Yes, we're going to spend that, but of course, they're not obliged to spend it. It's not like a legal requirement that they spend it. They can obviously reduce the speed of spending. That's what people forget. I mean, this CapEx stuff, they can just slow it down if actually the demand situation just isn't quite hitting those wild expectations, right? So it is variable, but people love to throw around the big numbers.

Yeah.

And can OpenAI afford $1.4 trillion? You know, well, they can't, but they're not going to spend that much.

Yeah. Such a good point. You know, you're kind of predicting a spend out from 5 years and when you look at the speed of change in the last 12 months, making a prediction of that for 5 years when it's almost compounding on that curve of improvement.

I think like you said, but it is so important though because it's taking up an ever-growing portion of like, if you look at the US economy, it basically isn't growing if you take out data center CapEx spend. But when you Don't take it out. Well, it's such a monster investment, but actually growth's great. So really the health of the US economy is dependent on whether this data center buildout actually goes to plan in terms of expectations or not. I mean, it literally now, their economic growth situation is really heavily dependent on it.

Okay. Well, let's quickly wrap through some of the rest that they have. VCs like Sequoia, for example, argue today's valuations are supported by real demand and revenue scaling across the stack. And maybe that wraps into where some of the opportunity sits. In terms of the stack reference there.

So, yeah.

Yeah.

That's what we were talking about before, about in a podcast a few weeks back, I think, where you've kind of got the first phase, which is the infrastructure and the chips, right? And that's why NVIDIA has been at the front center of this. Then you've got the kind of the, the sort of the AI platforms, like your, your kind of OpenAIs and your, and so on. And then you've got the long-term thing, which is like the application layer, which then enables companies to increase margins by reducing costs because they're more efficient now with these AI tools, right? So if you are positioned as a business through the stack, then you can kind of monetize more than one layer. Well then, you know, that's when it looks super interesting.

Yeah. And one of the things I did take from the report was someone who was quite bearish about this actually, and it was a New York University, NYU professor, and as an AI critic. So I think it's fairly well known, Gary Marcus. So just unpacking what he says and why he terms it quote, autocomplete on steroids. Which is the reliability problem that caps the real-world impact of generative AI. And what he's basically saying is that LLMs, so large language models, what they do, they generate plausible text based on patterns, not necessarily meaning, hence his statement. So they're very good at sounding smart, but basically they're just predicting the next word. So not reasoning like a human. And that leads to structural limitations, especially in high-stakes, real-world applications is his perspective. What do you think of that?

Yeah, I mean, look, he's got a point. I do think people probably, the rate at which things have happened so far has been so rapid with NVIDIA, right? I think people's expectations are a bit inflated about how quickly you'll get to that final layer of businesses genuinely seeing margin improvement through the use of these tools. I think right now everyone just needs to be seen to be using AI, right? Shareholders demand it because, oh, if you just use AI, we'll make loads of money. But the, you know, the kind of change programs going on internally to try and use these, the training that needs to go in, training humans to use them. And actually these tools, well, yeah, they're still brand new, right? So they're not You know, it's not AGI. So maybe really that's probably what your Gary Marcus is talking about here. It's like people getting way too excited about AI and really it's not until we get to the AGI phase that you're going to see all that sort of genuine kind of economy-wide efficiencies. Coming through.

So that structural limitation, he— 3 other things he highlights are the requirement of massive curated training data, which can still reflect biases and outdated assumptions. And then the fact that a lot of these models, they're not real-time models. So once they're trained, they're static until they're retrained. Hence, now I think I just saw my phone this morning, actually. It's covertly gone. I must have missed the email from ChatGPT 5.0 to 5.1.

Oh, is that right?

But there you go. It's just retrained because what's happened, I think 5 came out, what, 12 weeks ago? So just think of the data that's happened, but again, it's not real time when it's doing it. And then the black box problem, which is it's often unclear why a particular decision or generated output has happened. So, does that limit trust and therefore adoption when it comes to large-scale real-world cases? So overall then, Jack Goleman is saying the AI may not be in a bubble today, but they do sort of flag the ecosystem becoming circular, which we've talked about before. Things to watch being, you know, the debt part of the equation. So how debt-fueled does it become for this activity that we're seeing? On spending. And then maybe that last point, final words, the increasing dependency on a perfect execution.

Yeah. Well, that's it. The valuations are so high, they're pricing perfection. They are pricing this future that is, you know, that we arrive at without any hurdles or stumbles. You know, it's perfect company execution, but It's impossible. This is such a massive secular shift. It's impossible to have perfect execution. So we are going to get bumps. You know, when I like, maybe I think Goldman's are probably right. It's not a bubble. That doesn't mean we're not going to get some down corrections, right? And I think maybe we might see a little bit of softening as we go through Christmas into New Year. I do think we might get a bit of a pullback. But so I think they're right. And I think that's what they're saying. It's not a bubble, but it's not just straight up into the right clean line with no buts.

I do have a bit of a beef with that statement of the bubble because it's like, well, I don't think you're asking the right question in terms of, was the internet a bubble, like dot-com bubble? And you're like, well, look where we are today. From 1999, you're like, was it a bubble? You're like, so it's like the direction of travel is still, AI is going to be absolutely integral to every part of the economy going forward in the long term. So like you said, short term, long term.

But there is a difference though, the dot-com bubble, the NASDAQ index, So the average of all of the biggest companies in that space dropped like 90% and a whole chunk of them went bust. Now, obviously there were winners in the end. You know, you think about the Googles and the Amazons, right? Obviously long-term, well, now they're the biggest companies on the planet, so Obviously the internet boom was a thing and it did become real. So is this a bubble this time? It's only a bubble if the NASDAQ drops 90% because it bursts and everyone gets killed. Fine. You're going to get Nvidia still in business in 10 years' time and probably still the biggest company on the planet, right? But yeah, it's just whether or not we get a 90% market cap collapse. And Goblins are saying that's not going to happen.

I don't think I can refinance the home quick enough if that happened.

All right.

Well, thank you very much for listening. Yeah, you may have recognized if you stayed to the end that we didn't cover an M&A deal this time round, but fear not, that's because we are going to segregate the content as we were doing previously when Stephen was with us. And so now on Monday, you will get M&A designated conversation. So remember to subscribe to the channel, look out for that. So we'll deep dive into some deals on Monday morning. All right. Have a good, great weekend, everyone.