OpenAI Declares the Next Phase of AI

Today on the AI Daily Brief: some big questions around the next phase of AI and whether what we now call AI is actually multiple things. Before that, in the headlines: OpenAI joins the IPO filing party. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. Alright friends, quick announcements before we dive in. First of all, thank you to today's sponsors KPMG, Robots and Pencils, ZenCoder, and OutSystems. To get an ad-free version of the show, go to patreon.com/aidailybrief, or you can subscribe on Apple Podcasts. Remember, it is just $3 a month for ad-free. And if you want to learn more about sponsoring the show, send us a note at sponsors@aidailybrief.ai. Also, a big thank you to everyone who has filled out our AI DB for Teams survey. One of the things that's becoming very clear is that something that people are really looking for is an easier way to share specific nuggets from the episodes. We've got a couple new ideas for that coming up, potentially with a brand new webpage. So keep your eyes peeled for that over the next couple of weeks. For now though, let's jump into the headlines. You can absolutely feel things heating up right now. We'll have some dimensions of that in our discussion in our main episode, but even throughout the headlines, the stakes are just getting raised. Markets are firing up, energy is high. The whole AI space is just vibrating on a different level. One example of that, OpenAI has officially thrown their hat in the ring and filed to go public. The company filed their IPO paperwork confidentially on Monday. Now, as I mentioned after the Anthropic filing last week, confidential filing is the industry standard and certainly doesn't carry any significant implications. All it really means is that we don't get to look at the company's financials until much later in the process. We also won't know the valuation sought by OpenAI until much closer to the listing. And while the press is casting this as an all-out race between OpenAI and Anthropic, the quotes don't suggest a a lot of urgency from the companies themselves. When asked about the IPO last week, Anthropic President Daniela Amodei simply stated that the filing, quote, gives us the option to potentially go public after the SEC review. In an X post on Monday, OpenAI said, we have not decided on timing yet. It may be a while because there are things we want to do that are likely easier as a private company, but it's a complicated set of trade-offs, and this gives us the option to go public sooner if that ends up being best. Now, if you did decide that that was all just bluster and these companies were in fact in an all-out race, Judging from the speed of the SpaceX IPO, you're probably looking at September as the earliest reasonable possibility, but who knows, when it comes to AI, a lot of norms are getting thrown out the window. Now broadly speaking, I would say that there are two categories of takes when it comes to this IPO slate. The first is that the sequencing matters. ChubbyonX writes, going first could matter. The first major frontier AI IPO may define public market expectations for the entire sector, while later entrants risk being judged against that benchmark. And while that's certainly true, I find myself much more in the camp shared by Lassan on X who writes, I have a prediction for Anthropic, OpenAI, and SpaceX IPO. They will all go vertical. Or as the Kobayashi letter put it, we're about to witness the most incredible IPO run in history. Now as the SpaceX IPO approaches, Elon Musk has unveiled further plans for putting data centers in space. In a half-hour-long video posted to X on Monday, Musk revealed a prototype design for the data center satellites. Each satellite will be designed to handle around 150 kilowatts of AI compute, roughly equivalent to a single rack of NVIDIA Blackwells. Musk said during the presentation, "An AI satellite is essentially a lot of solar cells, a radiator, and you still need some laser links, but you don't have all of the super complex antennas that you have on a Starlink satellite. Given the two, the easier one to design for is the AI satellite." He later added, "Part of what I want to convey here is that there's not some magic that's necessary that doesn't exist for AI satellites. A lot of this technology we've already made for Starlink V3 satellites." We don't think this is a super hard problem compared to things we already do. Now when many people first heard Elon start to talk about data centers in space, there was a natural knee-jerk skepticism. And for people that dug in a little bit more, even outside of any pre-existing biases around Elon, there did seem to be some big, possibly insurmountable problems. Chief among them was heat dissipation, as the lack of atmosphere makes passive cooling impossible. We're now getting some details and it sounds like the SpaceX satellites will angle a thin edge towards the sun to minimize solar heating and use radiation panels to expel heat from the GPUs. This system is similar to the cooling already in use on Starlink satellites. Overall, the whole concept is starting to get a lot more credible in recent months, with Google reportedly in talks to partner with SpaceX for their own data center satellites. In other words, while there are still difficult engineering challenges, Musk is increasingly presenting a feasible pathway. His presentation also discussed the manufacturing side of the project. SpaceX plans to expand their Starlink facility to reach 11 million square feet, which would be around the same footprint as the Tesla Gigafactory in Austin. The facility, dubbed Gigasat, will be used to manufacture the gigantic solar panels that are required to power these satellites. Elon also discussed the scale and time horizon for the project. He said that SpaceX will try to achieve an annualized rate of putting 1 gigawatt of capacity in space by the end of 2027. That would require almost 7,000 satellite launches a year if they remain at 150 kilowatts each. After 2027, Musk said that SpaceX will attempt to scale capacity by an order of magnitude each year, eventually reaching a terawatt of capacity. Now that is a more ambitious schedule than disclosed in the IPO filing, and even Elon himself has acknowledged that it's all a bit speculative. He commented that the plans should be taken with a grain of salt. Still, the thesis of putting data centers in space is increasingly central to the SpaceX public listing. In their prospectus, SpaceX claimed that space data centers could be a $23 trillion market, which will to some investors still sound like he's saying that their market cap potential is approximately a bajillion dollars, is getting a little bit more credible by the week. Data center skeptic and hedge fundie David Orr kinda represents the current smartest guy in the room thinking, in that while previously he was ranting about how infeasible this was, he's starting to come around not to the idea that it's obviously going to work, but that now at least Elon is making a bunch of verifiable claims and getting into the details on how it's supposed to work, making it way more tangible and seemingly a lot more real. Orr wrote, "If SpaceX has an 80% gross margin on launches, they're a lot closer than I thought for data centers in space." It's still far off, but not wildly far off. I guess if Starship can bring this down another 80%, it could be feasible. I'm biased to thinking this kind of thing is BS, and it usually is, but maybe keeping an open mind here is right. Meanwhile, for the SpaceX IPO itself, demand seems red hot. Sources told Reuters that stock is roughly 2x oversubscribed, with $150 billion worth of orders for $75 billion in available stock. Now this level of demand wouldn't necessarily be out of the ordinary for a regular frothy IPO, What makes it impressive is the sheer magnitude, given that this is the largest IPO in history. One hedge fund manager said that there's career risk in abstaining from the sale, commenting, lots of people will have to explain why they don't own it rather than justifying a decision to buy it. Reports are circulating that brokers are being told to spread allocations widely across retail investors. The IPO has an unusually high 30% allocation to retail, and according to the chatter, brokers are being forced to ensure that small accounts still get their fair share. Analyst Phil Truby wrote, Elon is serious when he said retail will get allocations. In addition, some brokerages are telling traders that they need to hold on to their shares for 30 days or be banned from buying other IPOs later this year. Trades are being told to fund their accounts by the Wednesday close, implying that the first day of trading is indeed expected on Friday. Staying on infrastructure for a minute, chipmakers are turning to Intel as supply chain issues worsen. The Information reports that both Google and Nvidia are quietly adding Intel as a backup manufacturer for their AI chips. Until now, both companies had used TSMC as their single supplier for high-end chips, but of course, recently TSMC's order book has ballooned, leading to a multi-year waiting list. During a call with investors last week, TSMC CEO CC Wei warned that even with new fabs coming online in Arizona and Taipei, it'll be a long time before they can meet full customer demand. That kinda leaves Intel as the only viable alternative. Google has reportedly placed an order for 3 million TPUs to be manufactured in 2028 after being satisfied with test units. That's around half of the units Google expects to deliver across '27 and '28, according to estimates from Morgan Stanley. It's also the first major chip order for Intel during the AI era. Nvidia, meanwhile, is said to still be in the testing phase. Sources said that they're currently testing Intel equipment for producing the next-generation Feynman chip, set for production in 2028. Now, this is all big news for Intel, who have struggled for years to catch up on AI chip manufacturing. After taking control of the company last year, new CEO Lip-Bu Tan announced that Intel would double down on contract manufacturing for other chipmakers, with their in-house AI chip family called Gaudi having been a sales disaster. Faster. And while of course Intel have made big strides in their reclamation project, this story really speaks to how severe chip supply chain issues are becoming. There's no hint of dissatisfaction with TSMC here, or even a desire for redundancy. This appears to be purely about TSMC being completely out of spare capacity for the foreseeable future. Now investors had already been buying the Intel redemption arc, almost tripling the stock price in April before entering a downturn in May, and this news broke the month-long downtrend with the stock jumping by 11% on Monday. By the way, one more small one for my market folks out there. Goldman Sachs and JP Morgan are looking at trading compute futures as a way to hedge their data center exposure. Compute futures are expected to launch later this year and will allow traders to bet on the future price of GPU rentals in a similar way to other commodities like oil or wheat. The information reports that Goldman and JP Morgan are exploring the use of this nascent market as a way to hedge against the risk of data center overbuilding. And while many might read the introduction of compute futures as yet another speculative market, for clients that are heavily invested in the data center buildout, There's currently no direct way to hedge that risk, meaning that this sort of compute market can actually be useful outside of just speculation. Lastly today, it is not just in markets that we feel things heating up. Washington is taking another look at AI regulation with multiple agendas heading to Congress. Axios reports that the White House is negotiating federal preemption of state AI laws in exchange for supporting other key tech priorities in an upcoming bill. Preemption was part of the bipartisan bill proposed by Representatives Obernolte and Trahan last week, But it seems a Senate version is also in the works. Axios said negotiations are currently being led by Republican Senator Marsha Blackburn. Blackburn scuttled AI legislation last summer over the preemption issue, pointing to concerns about losing child and copyright protections that are already legislated in various states. This time around, Blackburn is looking to bundle those issues into a federal regulatory package. Her office said, Senator Blackburn is spearheading the negotiation with the White House to finalize legislative text of an AI preemption package that includes protection for kids, creators, and communities through the Senate version of the Kids Online Safety Act, the No Fakes Act, and age verification requirements. They also added that child safety would receive a carve-out from preemption. Sources told Axios that leading AI labs would be attending a White House meeting this week to discuss appropriate benchmarking for government vetting pursuant to the recent AI executive order. Now, this could be the actual purpose for the meeting President Trump discussed last week in the context of AI companies ceding shares to a public wealth fund. Separately, Democrat Senator Adam Schiff has introduced a bill to restrict the Pentagon's AI use. The bill would require the Pentagon to keep a human in the loop for autonomous weaponry, and protect against the use of AI for domestic surveillance. The bill then essentially legislates Anthropic's red lines which became the subject of their fallout with the Pentagon back in March. A similar bill was put forward recently by Senators Kelly, Gillibrand, and Slotkin, but Schiff is pushing to have his bill attached to the must-pass defense funding bill expected to come to a vote later this month. More broadly, these particular restrictions seem to be a core pillar of the AI platform for centrist Democrats heading into midterms and beyond. Schiff told the Wall Street Journal, we're no longer anticipating these impacts. They're here. AI could very well be the dominant issue for the next presidential election. Now, Schiff's bill is of course just one of a myriad of proposals coming from the Democrat side of Congress. Bernie Sanders recently presented his 50% tax on AI equity as legislation, while others are pushing for other forms of AI taxation. Michigan Senate candidate Marie McMorrow, who is running on a token tax applied across the economy, said, it feels like we are hitting a cultural tipping point. For the AI companies, meanwhile, it seems like regulation is inevitable, spurring far more engagement from lobbyists. OpenAI's Chief Global Affairs Officer Chris LeHane said, "We think we're going to need to even do more as we go forward given the speed that this is moving." Now, as you sit back and look at all of this, you could be forgiven for thinking that we are on the precipice of yet another phase, you might say, in AI. And interestingly enough, that's exactly what OpenAI has recently said, and that is the subject that we turn to in our main episode. One of the most important AI questions right now isn't who's using AI, it's who's using it well. KPMG and the University of Texas at Austin just analyzed 1.4 million real workplace AI interactions and found something surprising. The highest impact users aren't better prompt engineers. They treat AI like a reasoning partner. They frame problems, guide thinking, iterate, and push for better answers. And the good news? These behaviors are teachable at scale. If you're trying to move from AI access to real capability, KPMG's research on sophisticated AI collaboration is worth your time. Learn more at kpmg.com/us/sophisticated. That's kpmg.com/us/sophisticated. One thing I keep seeing in enterprise AI companies hedging across every cloud, every model, every framework, or paying a GSI for a pilot that never ends. The teams actually shipping, they've picked a lane and they move fast. That's one of the reasons I like today's sponsor, Robots and Pencils. They've gone all in on AWS. They're an advanced tier and AWS pattern partner, and they ship production AI coworkers in 45 days. That's led to them doing some of the more interesting work I've seen on AI coworkers. And by that, I'm not talking about chatbots. I'm talking about actual agentic systems that sit inside a business architecture and do real work. That kind of focus matters if you're an enterprise leader trying to get something real into production or an AWS rep trying to move a customer from interested to deployed. Request an AI briefing at robotsandpencils.com. One conversation with Robots and Pencils and you'll know. So coding agents are basically solved at this point. They're incredible at writing code, but here's the thing nobody talks about. Coding is maybe a quarter of an engineer's actual day. The rest is standups, stakeholder updates, meeting prep, chasing context across 6 different tools. And it's not just engineers. Sales spends more time assembling proposals than selling. Finance is manually chasing subscription requests. Marketing finds out what shipped 2 weeks after it merged. ZenCoder just launched ZenFlow Work. It takes their orchestration engine, the same one already powering coding agents, and connects it to your daily tools. Jira, Gmail, Google Docs, Linear, Calendar, Notion. It runs goal-driven workflows that actually finish. Your standup brief is written before you sit down. Review cycle coming up? It pulls 6 months of tickets and writes the prep doc. Now you might be thinking, didn't OpenClaw try to do this? It did, but it has come with a whole host of security and functional issues which can take a huge amount of time to resolve. ZenCoder took a different approach. SOC 2 Type 2 certified, curated integrations, tighter security perimeter, enterprise-grade from day one, model agnostic, and works from Slack or Telegram. Try it at zenflow.free. This episode of the AI Daily Brief is brought to you by OutSystems, a leading agentic systems platform built for the enterprise. Organizations all over the world are building, orchestrating, and governing agentic systems on the OutSystems platform, and with good reason. OutSystems' open and unified platform allows teams to architect, deliver, and scale governed agentic systems with agility. Teams of any size and technical depth can use OutSystems to build, deploy, and manage AI apps and agents quickly and cost-effectively without compromising reliability and security. With OutSystems, you can rapidly launch ideas from concept to completion. It's the leading agentic systems platform that is unified, agile, and enterprise-proven, allowing you to accelerate growth, reduce operational friction, and deliver real enterprise impact with AI. OutSystems. Build your agentic future. Welcome back to the AI Daily Brief. Today we're talking about OpenAI's blog post in which they declare something of a new phase in AI. Now, part of why this is interesting to look at is that it's very clear that we're living through a major transition in AI. On a micro level, we're in our second sort of transition in a couple months. The first, of course, being the agentic transition that happened between November of last year and January of this year. And the second being the shift from the token subsidy era, where we were consuming a lot more tokens than we were paying for, to the token shortage era, where the physical realities of the limits of the compute that we have available to serve means that state-of-the-art AI especially is getting more expensive. Now, when you take a farther step back, obviously these are parts of the same transition. And really, I think when the history books are written, the reason that this period feels so intense and immense is that it is the second massive transition in generative AI following the launch of ChatGPT itself. Now, I would argue when it comes to the ultimate shape of AI in society, this agentic transition and all the consequent economic changes that come with it is likely to be even more significant than the first. So much so that part of what I'm wondering today, as I look at the OpenAI blog post contrasted with the non-event that was Apple's Siri announcements at WWDC, if there is actually a fairly fundamental break occurring in what we now lump together as AI. But first of all, let's read some excerpts from the piece that they published, as it's not particularly long. The piece is called Built to Benefit Everyone: Our Plan. The first part of it is sort of your standard historical analogy. This is big sort of rhetoric. Every few generations they start, a new technology changes everything. Imagine electricity reaching a rural American town in the 1920s. Before power lines arrived, daily life was shaped by physical limits: hauling water, washing clothes by hand, preserving food with ice, and ending much of the day when the sun went down. Electricity did not transform every household overnight, and many of its benefits reached people unevenly. But as access spread, ordinary life changed. Light at night extended the day. Electric pumps, appliances, and refrigeration reduced some of the hardest daily work. Radios brought news, music, and connection from hundreds of miles away into homes and community spaces. The first promise of electricity was practical, but its deeper impact came from the new possibilities it opened up as more people could use it. With time, a lot of new possibilities emerged, with machines and computers greatly accelerating progress in medicine, engineering, and many other fields. By the end of the 20th century, the average lifespan had increased by over 20 years, and the median inflation-adjusted income tripled or so. These gains were driven in no small part due to the advances in healthcare, sanitation, and living standards, many of which were enabled or accelerated by widespread electrification and related technological progress. This, they argue, is happening again with AI. AI, they write, will soon be capable of extraordinary things, but the point is not the technology by itself, the point is what people can do with it. The positive impacts, however, they write, are not a foregone conclusion, or as they put it, the future will not happen automatically. They do the standard, there's going to be a lot of great things, but also clear out about the risks kind of paragraphs, with one interesting novel tweak on that theme, where you see the continued backing up from the narrative of full knowledge worker replacement when they write, "Entirely automating everything is not the future we want. It would be unfulfilling and it would be dangerous. AI should help people pursue their goals, not become untethered from them. As AI systems become more capable, the human role becomes more important: setting direction, making trade-offs, applying judgment, and bringing values, taste, care, and responsibility to work." Now, as a little aside, the way that I feel about OpenAI finally coming to this rhetoric and this idea, that AI in fact will not be used just to replace people, but instead to unlock their dreams in ways that were never imagined before, I'm reminded of a scene in The West Wing where press secretary CJ Cregg reveals to her communications colleague Sam Seaborne that she doesn't understand the census, even though she's been acting as she does for weeks. After a little pause, Sam Seaborne, played famously by Rob Lowe, says, "I tell you what, let's forget the fact that you're coming a little late to the party and embrace the fact that you showed up at all." In any case, from that new rhetorical shift, they move into what is the actual substance of the post. They write, "We believe that AI, doing AI research, will become the determining factor of the pace of progress within the next few years. Faster technical progress makes human judgment and public coordination more important, not less. The future should be shaped by people, institutions, and societies, not only the companies building the most capable systems." Indeed, reflecting the broader conversations at the moment, they write that they expect national and global coordination to become more important, arguing that they have, quote, "long believed there should ultimately be an international organization that helps coordinate leading AI efforts." to reduce catastrophic risk. And then we come to the core internal-facing part of the post: OpenAI's 3 current main goals. Their first goal is build an automated AI researcher. An AI system, they write, that can accelerate and increasingly automate the research process itself while remaining steerable, accountable, and connected to people. Our internal belief is that by March of 2028, we may have a significant fraction of our research being done by AI systems in tandem with our own researchers. To make sufficient progress on alignment, we believe we will need AIs to iterate alongside us. This will help us navigate the transition to the post-AGI world so that we collectively decide the path towards the future. Goal 2: Accelerate the economy by accelerating scientific progress, productivity, and economic growth while working to ensure the gains are widely shared. Everyone should have an opportunity for a meaningful share in the prosperity AI creates. 3: Give everyone on Earth a personal AGI, empowering them to benefit from one of humanity's most transformative technologies in whatever way they choose. And thus they say, to be able to deliver on this, we are entering the third phase of OpenAI. The first phase of OpenAI was about doing research toward AGI. The second phase began when our research became relevant to the real world and we became a product company, deploying our systems, learning from how people use them, and making continued progress toward AGI that is safe and aligned with our mission. Now, they declare, we are entering the third phase. The economy is beginning to reshape around AI. The central question now is how to make advanced AI abundant, affordable, safe, useful, and easy enough for every person and organization to benefit from it. Frontier capability is only part of the job. The bigger task is turning that capability into tools people can actually use to thrive. Above all, we believe a broad distribution of power will help lead to a better future. Human history shows that concentrated power creates fragility, while widely shared power makes societies more resilient, adaptable, and free. A good AI future cannot be one where a small number of institutions control most of the capability and most of the upside. It should be a future where many people, companies, communities, and countries can build, benefit, and hold power. We believe this transformation should belong to everyone. So that's the post. And when it comes to interpretations and the question of why this was released now, for many, it was impossible to separate from their confidential IPO filing, which happened earlier in the day and which we just talked about in the headlines part of the episode. Gennaro writes, Sam Altman just called OpenAI's 3-phase plan built to benefit everyone on the same day they filed their S-1. Headlines: altruistic mission statement. Uh, it isn't. This isn't a roadmap. It's market segmentation. Consumers buy the dream, investors buy the TAM, regulators buy the public benefit corporation. There are infinite versions of this take, and all I will say is that for the next 4 months or so, until the IPOs for Anthropic and OpenAI are both in the rearview, you're going to have people frankly reasonably argue that every single thing that they're doing is in some way, shape, or form about getting the optimal outcome out of that IPO. Now, for others, the big emotional takeaway was certainly the idea of power distribution. Teng Yan writes, TL;DR: OpenAI's Phase 3 is that AI should be for everyone, not concentrated in a few hands. Stanford prof Andy Hall writes, "Concentration of power seems to be the central political economy question of AGI." And certainly given the rising discourse around Bernie and sovereign wealth funds and Trump taking a cut, the discussion at least around how to ensure AI benefits everyone is reaching a crescendo in a way that it never has before. Now for others, This is all about reading the tea leaves of what's going on behind the scenes, and a speculation that the labs now think that we're beyond some event horizon based on things they've seen that aren't available to the public. Chubby writes, sounds like we're now taking the final steps towards AGI and post-AGI. Prinz writes, conspiracy theory time, OpenAI just announced that we're entering the third phase. Why today? Why now? Blog post was co-authored by Sam Altman and Jacob Pachocki, who we know is in charge of automating AI R&D at OpenAI. You may remember that the livestream where OpenAI's plans to automate AI R&D were announced was also Sam and Jacob. No mention of the goal to produce an intern anywhere in the blog post. The goal is now just the automated AI researcher by March 2028. All weekend long meanwhile, the Codex team posted about loops. Sam Altman posted about recursive loops last night. Do they have it? And by it, I think Prinz is referring to some version of recursive self-improvement or RSI. Lassane argues no, of course not., but they could have a much larger and stronger model. Now, the other event that happened yesterday that a couple of years ago would have been almost dead-on assured to be the lead story of the main episode was Apple's WWDC event where they announced that Siri AI was finally coming. Now, you might remember that we first started hearing about the new quote-unquote Siri AI back in 2024 when Apple launched, in huge air quotes, Apple Intelligence. The Verge writes, the whole intelligence bit of the Siri redesign was coming soon, Apple promised, It didn't. In fact, its promotion around Apple Intelligence was so misleading that the company is settling a class action lawsuit and has to pay iPhone owners for the features it never shipped. But now the new Siri is finally here, and basically it does the things that it was supposed to do the last time around. It can summarize your messages, add an event to your calendar, search the web. Basically, for the first time, Siri promises to do more than just set a timer incorrectly. And for some observers, especially those outside of the hyper-focused AI discourse, —this was just enough to be fine. Bloomberg's Apple watcher Mark Gurman wrote, "While there was nothing revolutionary, Apple just rebooted the foundations of its platform with functional AI, a working Siri, and improved performance. This is critical ahead of the next 3 years of blockbuster new devices that run these operating systems." The right move. Or as Samsung's David Lee put it, "Didn't try to win the game with 1 possession when down by 10." IDC analyst Francisco Geronimo said in a note, Apple does not need to win AI by having the biggest model or the loudest demo. It needs to make AI trusted, useful, and invisible across the ecosystem. And yet, of course, on the flip side, people who have spent the last 6 months playing with OpenClaw and becoming Codex and Claude Code power users look at this and are just absolutely baffled. Tenebris writes, "New Siri are fine, looks like it will be a legit useful upgrade and I'll make use of agentic search over my own text messages. Upgraded voice transcription might kill a few startups, but it's also just exactly the minimal stuff they should have launched in 2024." Signal put it more bluntly, Siri is basically ChatGPT 1.0 that has no impact on our work. Now on top of that, you also got some articles trying to argue why sneakily Apple is actually going to disrupt ChatGPT. Michael Malowicz wrote, Apple just killed paying for AI. In it, he basically argues that Siri in your phone, while it isn't going to be managing an agent swarm or refactoring your COBOL codebase over the course of a week, can probably order you a burrito about as well as Opus or GPT-5.5 can. And for a lot of people, that's all that's going to matter. And here's what hits me like a truck. I wonder if we have officially hit the point where considering consumer AI, i.e., the day in, day out turning to a chatbot to help answer a question or find some information as part and parcel of the same thing as work AI, where we are literally managing fleets of new synthetic employees that can unlock capabilities that were never possible for us before, is just fundamentally wrong. Now, nominally, all of these platforms have all of these features. You can use Claude AI, you can use Claude Code, you can use ChatGPT, you can use Codex. And with Gemini and Google, it's even blurrier. And yet, in some ways, I think this might have distracted us from the fact that we might be at more of a fork than these companies are willing to admit. It is so clear watching OpenAI's moves that the seat-based revenue of ChatGPT is so ridiculously inconsequential compared to the API usage of Codex that I wonder if on some level ChatGPT itself is becoming a distraction. Now, certainly they're giving some interesting signals to that effect, talking about blending the ChatGPT and Codex apps, building a super app. And yet still most of the discourse is talking about this super app as though it's a thing for consumers. Now, I think that there's no way, even if OpenAI has determined that ChatGPT is kind of a distraction and one of those side quests that Fiji Simo said that they shouldn't be spending their time on, that they'd actually make moves to kill it. First of all, they're going to see it forever as a great top of funnel for getting people into the Codex ecosystem. But second, it is a massive differentiator between them and Anthropic as they go public, and they're going to need to highlight those differences when it comes to the public markets. But I genuinely wonder if they actually care about consumer AI anymore. Even the way they're talking about AI benefiting everyone in this blog post, it's one about the financial and deterministic and control benefits, benefits and participation, and two, by saying that we're giving everyone a personal AGI, almost dragging them into those work-related use cases. As opposed to more traditional consumer AI stuff. Now I sort of wonder what it would change if we were allowed to care less about consumer AI. Would it mean less shoving it down people's throats with every app, including non-work apps, racing to add their AI features? Would the fact that this was a work technology, not an all-encompassing consumer meme, actually make people less angry? I don't know, but it's interesting to think about. I would not go so far as to say that consumer AI has failed. One need only look at how totally embedded in people's lives chatbots like ChatGPT have already become to know that that's not the case. But at the same time, it is also clear that there is no comparison between that and the sheer tonnage of impact that agentic work-related AI and the consequent business models and infrastructure are going to have on the face of society. And maybe it's time we start talking about them as separate things. Anyway, if the goal of OpenAI's piece was to get people talking, mission accomplished. For now, that's gonna do it for today's AI Daily Brief. Appreciate you listening or watching. As always, until next time, peace.