I see a large IPO on the horizon.
You're surrounded by jungle. Hold your position.
That's misinformation.
According to clearing order, let's just roll.
We are surrounded by jungle. Hold your position. Come on. Get up.
Trust the experts.
We are founder.
I see multiple journalists on the horizon. Stand by.
UAV online. Blaze. Double blaze. Triple blaze. Double kill.
Michael is w team deathmatch.
We are experts. Triple blade. That's just wrong. Right.
Hockey clearing order inbound. Get up.
We are surrounded by journalists. Hold your position. Strike one. Strike two.
Activate golden retriever mode.
Mark clearing order inbound. Five cook founder.
You're watching We Are Not a Car. We are not a car. We are not a car. It is Thursday, April 16th, 2026, and we are live from the TBPN Ultradome, the temple of technology, the fortress of finance, the capital of cars.
The capital of cars. Absolutely wild podcast between Nvidia's Jensen Huang and Dwarkesh Patel. So many clips, so much debate. We're going to kick it off with the question of whether or not Nvidia is a car. First, let me tell you about our lineup because we have an insane group of guests joining us today.
Kicking it off with Ben Smith from Semaphore. I think it's his third time on the show. He's live from the Semaphore World Economy Summit. He has convened a massive group of policymakers, CEOs, investors. He's going to give us the update on what conversations are happening on the ground. Then Jonathan Criss is coming on to talk about Starlink for Water. Former SpaceX employee just raised $24 million to enable off-grid desalination. One of my favorite topics these days. We talked a little bit about it with Peter Diamandis. Then Paul Scherer from IEN raised $15 million from Benchmark. James Morrissey and Kevin Hart are coming on to talk about building a consumer social play. Yeah, Kevin Hart's coming on. We will be asking him about Jensen. We've got to ask. I got to know from Factory Ulisses. We have a massive $46 million fund raise today for underwater and surface drones. Charlie Cheever from Expo is coming on. And Victor from Slash is coming on as well. And then we'll close it out with Theodor from Cognition. So, big show, should be going at least two and a half, three hours as usual. We will be off tomorrow, but we'll be back on Monday. Let me start off with from SemiAnalysis because it is a treat.
Wait and pause for a second. I don't... You know why I was asking you? Like what app you use to edit videos? It was to make this exact video and then I saw this and I was like, oh, I don't need to make it.
Yeah. You're not talking to somebody who woke up a loser. And that loser attitude, that loser premise makes no sense to me. We are not a car. We are not a car.
Such a great edit. It's absolutely such a good vibe.
And it's so funny because... I love Tokyo Drift. We're good.
It's funny because it's this new type of vibe where the SemiAnalysis team obviously they recontextualized the two clips and added the beat to get the video going, but that final edit with the cars sort of morphing together like it goes from there's a match cut. These are match cuts between one stick shift move and another or one car drifting and then another car drifting. That takes a really long time. Somebody clearly made this video just because they're an enthusiast of Fast and Furious and then the SemiAnalysis team was able to quickly recontextualize that to be about Jensen's answer on Nvidia's moat and the CUDA moat and what's happening with Nvidia. So let's go through this question because it's been bubbling for a while and it's sort of bubbled up most recently because there was a whole bunch of news that Mythos might have been trained on Trainium, then maybe TPU, then maybe Blackwell, and sort of a mix. And it just feels like more and more of the AI labs are capable of making other chipsets work. In the early days, it was all about Nvidia and now it feels like the incentives are really high to figure out other options and that creates a different competitive dynamic. So we can run through my thoughts on this and then we can go into the geopolitical considerations as well because that was another fascinating segment of the interview. So, Jensen, the CEO of Nvidia, spent over 90 minutes, it was almost two hours when you include the ads too, in the ring with Dwarkesh Patel. It was electric and the key question at least for the start was whether or not Nvidia is a car. And I'm only half joking about that. It sort of was the key question. Nvidia has been the market leader for years. During the gaming boom, Nvidia was the gold standard for rendering PC graphics. There was always decent competition from AMD. But once the AI boom kicked off, the CUDA ecosystem significantly sped up development of AI systems and training of AI models. And for those who aren't familiar, CUDA is a programming model that enables GPUs to accelerate demanding workloads by parallelizing computation. So instead of needing to work on the very low-level instruction sets, if you are a CUDA kernel engineer, CUDA engineer, you can access all the power of the GPUs very efficiently while staying up in the more mathematical AI research, more standard Python, C++ programming paradigms. You don't need to dip down too low, but it's getting easier to dip down lower and that's what we're seeing. So that created the CUDA moat because developers were way more productive and the biggest bottleneck to progress was allowing AI researchers to quickly test ideas and scale their experiments up to whole fleets of GPUs and eventually entire data centers. So at the time, researchers really liked CUDA and really liked Nvidia and they did not want to have to spend hours and hours figuring out other hardware systems. They just wanted to run their tests and see if the model was getting better and continue to scale. But recently the biggest cost center for AI labs flipped from researcher time more or less to compute capacity and this creates a much larger economic incentive to figure out a way to drive down the cost of chips used to train AI models. I wrote about this back on Tuesday, October 22nd of last year. So almost six months ago. I said not every link in the supply chain of AI can be completely commoditized. Nvidia has an insane amount of power having ramped full year revenue over the last three years from $27 billion to $60 billion to $130 billion. Absolutely insane top line revenue ramp at that scale. And that's why Jensen is so confident about how dominant this business is. It's the world's biggest company for a reason. It has been growing spectacularly at immense scale. And not only did they grow the top line but net profit margin grew insanely. So it grew from 16% to...
You do that when you have pricing power and massive leverage because you know in this case demand massively outstripping supply plus developer love and kind of just like...
And I believe the forecast net margin was 70-something percent. We were hearing about 80% potentially. And so Nvidia, you know, the plan and the plan is still to make an incredible amount of profit off of these chips because they are incredibly valuable. But all the hyperscalers and the AI labs, they are sort of incentivized now to form a bit of an anti-Nvidia alliance to commoditize the accelerator market and drive down those margins at least a little bit. And so today the AI chip market is starting to look much less monopolistic. AI coding agents can make it easier to write software that works on non-CUDA chip stacks. And the teams behind competing chips have plenty of resources and economic incentives to bring performance in line with Nvidia. Even if it's going to be a big hassle, even if you're going to need to spin up a team to get AMD or TPU working, it's going to be worth it because you're talking about billions and billions of dollars spent on chips.
Yeah. Another example of like, you know, an instance where an AI lab had so much urgency that they were willing to spend whatever it took was Meta rebuilding like Llama.
Yeah. Exactly. Like they were willing to outspend pretty much any other lab on talent because they didn't have time to find homegrown talent or go through a normal recruiting process. Especially considering a lot of those engineers were like happy doing what they were doing.
Yeah. Yeah. Exactly. Yeah, the incentives flip when you get to this scale or the incentives just get so big you can build a whole team for a specific thing, solve any problem. And so for the past few years, Nvidia has sort of looked like SpaceX's launch program. It's an incredible technology with very few viable alternatives. And so that creates great margins as we've seen with SpaceX's launch capacity. And they control something like 90% of the launch market. While the products have not degraded, quite the opposite actually, Blackwell and Vera Rubin are incredibly powerful chips. They're clearly on the leading edge there. And they have an incredible amount of supply chain guarantees from TSMC and across memory and all the other different pieces of supply chain. Like Nvidia is ready to make more chips, but increased competition makes the category look a little bit more like the car market than the rocket launch market. And so that I think is where Jensen is pushing back and saying, 'No, there's a lot more that we bring to the table with our customers that don't think of us like a car. This is not, you know, the difference between a Ford and a Toyota. They all sort of get you to the same place and you can swap one out. You can be driving a BMW one day, go to the dealer, turn it in for a Mercedes, and you're going to have a pretty...
Yeah. The other example that I was using with the team earlier was this idea of like if you're a delivery company like FedEx and you have a lot of like Ford vans and then Hyundai comes to you and says, 'Hey, you've been spending like $50,000 per Ford van, but like would you consider a Hyundai van? We'll sell it to you for $35,000. It's just as good.' And it could be like mildly inconvenient for the company because like they're kind of used to using Ford. They maybe have like an internal team that does some maintenance. But when you start looking at that cost differential, it can start to get pretty interesting to say like, 'Hey, why don't we try out some Hyundai? Let's move over some of our routes to Hyundai's and see how that goes.' And they try it and they're like, 'Hey, this actually works pretty well. Maybe there's like, you know, higher maintenance, but it actually like maths out.' Yeah. And like we're going to actually start adding more Hyundai to our overall fleet. And so Jensen's point on the podcast is that like we're not selling cars, right? This is not like something like you can swap out and Dwarkesh was obviously pushing back.
And for a lot of companies swapping out Nvidia for TPU would be very difficult. Some workloads that Jensen focuses on. He says we're not a tensor processing unit. We're an accelerator. There's a whole bunch of scientific computing workloads that work particularly well with Nvidia. The problem that...
And of course was just saying like yeah well it's perfectly fine to have like a specialized chip for specialized workflows because the biggest companies in the world, the biggest buyers here have like a single type of workload that they're trying to do and that's why they're using your competitors.
Yeah. And so, because the AI buildout does not seem to be slowing down as long as power can continue to be brought online and data centers can continue to find towns that will approve of them, demand for chips will presumably grow. But every chip designer and AI lab has to be praying that those net margins come down. How quickly will it happen? Very early to tell. The market did not react negatively to this back and forth in the short term. Although the price of Nvidia has been basically flat since last August and I think that's why Jensen is trying to sort of like reset on the narrative because it's possible that, you know, there's a lot of movement up and down. They've been sort of flat. There's a desire to sort of like reset and recontextualize.
Yeah. Inference demand is scaling. Yeah. In a very significant way and the Nvidia stock price is relatively flat.
Yeah. And since August there's been so many different moments where fears of the AI bubble, fears of the products not finding product market fit, that revenue might stagnate. Like we've seen a ton of bullish signals for like AI demand broadly. Demand is there. And so if you're a supplier, you should also be going up, but there's been this overhang of what will happen to margins and market structure. And so that is what people are going back and forth on. Tyler, did you have any other thoughts on this?
Um, I mean like personally I think I probably am more on the Dwarkesh side where like, yeah, I mean at some point like these margins are so high, there's so much opportunity here. Like we've seen that like actually, you know, if you're a big lab you actually can just like put a, maybe, you know, it's a lot of resources but you actually can just like train a model on a different architecture. You can serve them on a different architecture. Like you actually can figure these things out. As models get better, you can, you know, you can go lower and lower on the stack. Yeah. You can, you know, you can write the kernels. Yeah. Semi-autonomously, these things get faster and faster. And it's like I'm probably much more on the Dwarkesh side.
Yeah. It'll be interesting to see how Grok fits into this. The new CPU integration between different pieces of the puzzle. And then also I mean Dwarkesh makes this point that the supply agreements that Nvidia has might be a bit of a moat for the next few years while TSMC line time is so constricted.
Yeah. But even then like, you know, Jensen was saying the kind of supply constraints, this is like a two to three year problem. After that like you can just solve these things. Yeah. So it's like almost like I don't know how to think about this because, you know, to me it seems like yeah so much of the value of Nvidia is just like they have such an incredible relationship with TSMC but if and it's so valuable because of how constrained it is. But if that kind of constraint is maybe going to, you know, go away to some extent at TSMC.
Yeah, that's what he says like that they're going to increase, you know, they can build a new fab in whatever, three years. I mean the big takeaway from the conversation is like you have one person who seems incredibly AGI-pilled which is Dwarkesh and then you have Jensen who doesn't seem AGI-pilled in that sense at all, right? He... Dwarkesh asks about, you know, he was kind of getting at early on the idea of like can you just like prompt your way to Nvidia chips? He's like you basically sell software. And he didn't put it quite as aggressively. But yeah.
Yeah, that's of course like, you know, more on the nose, but Jensen was basically like no I don't think that'll happen. And then when you get to the whole geopolitical conversation too again it's like Dwarkesh is like you're selling nukes and Jensen is in my view like I'm selling computers. Yeah. Right. Yeah. And that was like the big rift. It was like these two kind of totally conflicting world views and it made for some very... pay-per-view. It was a pay-per-view. It should have been a paper.
Yeah. Shriram Krishnan summed it up pretty well. He said, 'Every person here's reaction to the Jensen plus Dwarkesh podcast can be extrapolated directly from whether they believe in the Frontier Labs achieving short timelines for AGI, ASI. If you believe in the labs achieving RSI and then AGI, ASI for some definition of all three in the next few years, you're probably sympathetic to the frame Dwarkesh adopts. If not, you're probably more sympathetic to the arguments from Jensen.' And so we can go into the export controls next and talk about that. Metacritic Capital sort of summed up a little bit of why just Jensen's rhetoric and how he wasn't conceding a lot of things. Dean Ball said, 'Dwarkesh reveals how inconsistent and unbattle-tested AI acceleration talking points are, especially when they're filtered through the prisms of corporate comms and mass politics. Strategically coherent accelerationism is possible,' he says, 'I try, but not currently prevalent.' And Metacritic Capital says, 'The problem is that Jensen doesn't concede anything. Compute spending going to the moon. One trillion revenue in sight. Models keep getting better. No unemployment. Software codes are good. Other Western accelerators are bad. Chinese competitors are good. Nvidia makes token costs decline 90% per year. But Chinese compute scientists are capable of making all the necessary algorithmic improvements. He also can't be AGI-pilled enough because at the end of the day, he is an intellectual property company in the business of sending a file to TSMC. I think it's part of Taiwanese culture to want to be loyal to all clients and don't have favorite winners. He doesn't want to betray his software customers. He has antitrust concerns.'
Yeah. He was making the bold case for software, which was that AI agents are going to use tools. So, he's like there's going to be more users of software than ever, which is something I'm like somewhat sympathetic to, but yeah, it definitely... it's still very easy to take the counter... the flip side of that. Yeah. Well let's play the distilled recap from Dwarkesh Patel of the back and forth with Jensen on export controls. It's about four minutes and we'll watch this and then discuss. So...
If Chinese companies and Chinese labs and the Chinese government had access to the AI chips to train a model like Claude Mythos with these cyber offensive capabilities and run millions of instances of it with more compute, the question is, oh, is that a threat to American companies, to American national security? First of all, Mythos was trained on fairly mundane capacity and a fairly mundane amount of it by an extraordinary company. And so the amount of capacity and the type of compute that it was trained on is abundantly available in China. And so you just have to first realize that chips exist in China. They manufacture 60% of the world's mainstream chips, maybe more. It's a very large industry for them. They have some of the world's greatest computer scientists, as you know. Most of the AI researchers in all of these AI labs, most of them are Chinese. They have 50% of the world's AI researchers. And so the question is if you're concerned about them, considering all the assets they already have, they have an abundance of energy. They have plenty of chips. They got most of the AI researchers. If you're worried about them, what is the best way to create a safe world? Well, victimizing them, turning them into an enemy, likely isn't the best answer. They are an adversary. We want the United States to win. But I think having a dialogue and having research dialogue is probably the safest thing to do. This is an area that is glaringly missing because of our current attitude about China as an adversary. It is essential that our AI researchers and their AI researchers are actually talking. It is essential that we try to both agree on what not to use the AI for. With respect to China, we want to have, of course we want the United States to have as much computing as possible. We're limited by energy. But, you know, we got a lot of people working on that and we had to not make energy a bottleneck for our country. But what we also want is we want to make sure that all the AI developers in the world are developing on the American tech stack and making the contributions, the advancements of AI, especially when it's open source, available to the American ecosystem. And it would be extremely foolish to create two ecosystems: the open source ecosystem and it only runs on the Chinese tech stack, and a closed ecosystem and that runs on the American tech stack. I think that would be a horrible outcome for the United States. I mean, I think the concern going back to the flop difference and the hacking is yes they have compute but there's some estimates that because they're at 7 nanometer, they don't have EUV because of chip making export controls, the amount of flops they're about to actually produce, they have like one-tenth the amount of flops that the US has. And so with that, could they train eventually a model like Mythos? Yes. But the question is because we have more flops, American labs are able to get to these level capabilities first because Anthropic got to it first. They say, 'Okay, we're going to hold on to it for a month while all these American companies, we give them access to it, they're going to patch up all their vulnerabilities and now we release it.' Furthermore, even if they trained a model like this, the ability to deploy it at scale, you know, if you had a cyber hacker, it's much more dangerous if they have a million of them versus a thousand of them. So that inference compute really matters a lot. In fact, the fact that they have so many researchers who are so good is the thing that makes it so scary because what is it that makes those engineer researchers more productive is compute. We should always be first and we should always have more. But in order for that outcome, for what you described to be true, you have to take it to the extremes. They have to have no compute. And if they have some compute, the question is how much is needed. The amount of compute they have in China is enormous. I mean, you're talking about the country is the second largest computing market in the world. If they want to deploy, aggregate their compute, they got plenty of compute to aggregate.
Very, very tense back and forth. Dean Ball says it's a shame Jensen mostly fails here because the monoculture on export controls is bad. If you're a young AI policy researcher trying to make a name for yourself, it's almost impossible to be taken seriously unless you are pro-export controls. Monocultures are usually bad. And I am sympathetic to Dwarkesh's points there for sure. Especially on the inference side. Even if models exist in both worlds, like having a whole bunch of, you know, good guy compute that can go and patch bugs while the amount of attackers is much smaller. It's just a matter of, you know, how many resources you have on each side. That's a great point. The only thing that they are like talking around is just Taiwan as a particular, you know, particular turning point and how their various positions flow through to Taiwan policy and where the, how the China stance on Taiwan is something that I've always puzzled and I wish that both of them had articulated their sort of philosophy on actually wargaming out what export controls do to likelihood of Taiwan intervention or blockade or anything like that. I don't exactly know. I've been trying to like work through it, but I don't have a complete thesis, but we've been debating it back and forth all day. I don't know if you have a strong take on any of this, Jordi.
I appreciate Matt Zitelland's point. I kind of appreciate that Jensen Huang seems relatively normal about non-business stuff compared to other tech founder CEO types, but then when it comes to Nvidia's actual operations, he's a complete sicko.
Yeah. Yeah. I saw a couple takes around this that he had like some very, very strong points if you're deeper in the supply chain. I couldn't really assess how he did on that, but some people were. There were definitely people that were in Jensen's camp. It was divided, which I think is why this went so viral. Elliot Arledge summed it up here. Jensen Huang on Dwarkesh today. Most combative interview he's done in a while. The biggest regret not funding.
Has there been a more combative interview ever with Jensen? It seems unlikely.
I think so. No. And I'm trying to think of just like in terms of combative, like people were sort of comping this to like the Elon Dwarkesh interview which wasn't combative but it was more just like digging into all the different projects and trying to elucidate the strategy and how they all fit together. There's actually some interesting updates on where XAI might be going with a partnership with Cursor that we can go into. There is some funny details in here. Apparently the Larry and Elon begged Jensen for GPUs at dinner story. That never happened. We absolutely had dinner. At no time did they beg for GPUs. Which is funny. I wonder what would happen. Quietly folded Grok into the CUDA ecosystem. Reason premium ASP tokens where latency beats throughput. Higher throughput used to always win. That trade-off has bent though, bent enough that Nvidia is expanding the Pareto frontier towards fast, expensive, low throughput. And it will be interesting to see how all of that comes together. On ASIC cost savings, Nvidia margin is 70%. ASIC margin is 65%. What are you really saving? Someone still pays Broadcom. Open challenge to Google and AWS. Publish TPU and Trainium on MLPerf and Inference Max. I would welcome Trainium to demonstrate their 40% claim. And so, Jensen threw down the gauntlet and said to Amazon and Google, head over to SemiAnalysis and put your chips on Inference Max so that they can be properly benchmarked so we can see.
Yeah, that was a very, very good call out.
Yeah. Yeah. It was smart. I just don't know why they don't go over and... I've been excited Inference Maxing to see the Inference Max results. Dwarkesh posted tomorrow and it's him and Jensen standing next to each other and Alex Vulkov said, 'Hey, Dwarkesh, was this picture taken before or after the pod?' Because it does feel like it was a tense situation. Although, you know, to both of their credit, like Jensen, it felt like he loved being in the arena, getting asked hard questions, like working through this. There's this back and forth where Jensen's pushing back and Dwarkesh says, 'Ah, I can drop it.' And he says, 'You don't need to drop it. I'm enjoying this. Let's hash this out.' And I thought that was very diplomatic and just good overall. Well, Intel is up on the news, up 4% today, 10% over the past five days. Almost at all-time highs. I think we're very close to the 2000 peak for Intel, which it was also around where they were trading in 2021. $67 a share, $330 billion company. Clearly with all of this backdrop and just the idea of more chips and maybe the CUDA ecosystem being something that you can work around, can an American fab run by Intel produce a chip that's viable for an AI lab? It feels like increasingly yes. That's certainly the argument that's being put forth by Dwarkesh and it would be very exciting. I think it's something that everyone would support an Intel resurgence and there's some news around TerraFab potentially getting involved. But first, let's start with the scoop from Grace K over at Business Insider. She says, 'Scoop, Cursor plans to use XAI's infrastructure to train its Composer 2.5 code.'
Where's the golden scoop, Tyler? According to people familiar with the matter, Cursor will use tens of thousands of XAI's GPUs. They said and we got a scoop for Grace K.
This one's going to Grace. Grace, congratulations. You win the golden scoop. The golden scoop. Congratulations.
Now interesting to see something we talked about probably midway through last year. XAI's, you know, shown a tremendous ability to, on the kind of infrastructure data center side, spinning up a huge amount of compute very, very quickly ahead of any timeline that any reasonable party would have probably expected. And demand hasn't exactly followed in the way that they would have liked and so opening that up to a company like Cursor who has all the demand and what they really need is their own model.
Yeah. It was also interesting because I don't know if it was thrown out as a potential project for other companies. I feel like MSL mentioned it at some point, maybe OpenAI, but there was some talk of like, okay, if you're marshalling all this compute and you wind up with too much, like what do you do then? And the idea of becoming a cloud provider if you have a data center and everything.
And that's been the big question of like, you know, with everything that SpaceX is doing and now TerraFab, they're going to be creating all of this capacity. Yep. But what is the, where's the demand for that capacity going to come from, right? And so you could imagine a world in the future where if SpaceX has a bunch of space data centers they open up that capacity to a bunch of companies other than Yep, just Elon Inc.
So, Grace says, 'The setup effectively turns XAI into a kind of cloud provider by renting some of its GPUs to other companies. XAI could start generating revenue from its massive infrastructure while still developing its own AI models. The arrangement could help the company offset the costs of building and operating data centers while also deepening ties with a startup that has access to valuable coding data.' And so, there could be some sort of trade deal going on. Ed Lello at Bloomberg has a report from the TerraFab. Musk's team is actively requesting price quotes and delivery timelines for a wide range of chipmaking equipment, photo masks, substrates, etchers, deposition, cleaning, testing tools according to sources. Elon Musk's lieutenants have reached out to chip industry suppliers for his envisioned TerraFab project. Remember he was pictured with Lip-Bu Tan from Intel I think last week. Early steps in an audacious and likely arduous attempt to break into the production of cutting edge chips. That is a very, very tall order. But maybe there's never been a better time to break into the cutting edge chip market given that you don't need to, I mean you sort of need to reinvent CUDA but it's becoming easier potentially. That is the story. Well there are two big releases from Anthropic and OpenAI today. Claude announced Claude Opus 4.7, our most capable Opus model yet. It handles long running tasks with more rigor, follows instructions more precisely and verifies its own outputs before reporting back. You can hand off your hardest work with less supervision, they say. Very good score on SWE-bench Pro, 64.3%. Excited to test it. They say Opus 4.7 has substantially better vision. It can see images at three times the resolution and produce higher quality interfaces, slides, and docs. As a result on the API, the new X high effort level between high and max gives you finer control over reasoning and latency on hard problems. I'm excited to see how these do on ARC-AGI V3.
I think the most notable thing here is you have a model card that shows a model that's not publicly accessible.
Yeah. Yeah. They share the Mythos benchmark. So Mythos mogging Opus 4.7 just sitting there, but of course unless you're one of select companies, you won't be getting access to it at least yet.
Well, Addi gave Opus 4.7 the tiny man bench, which is a very funny one where you ask the model, 'I am a tiny man. When my son was born, they handed me to him.' And Opus 4.7, that's a nice little riddle. The answer is a bar of soap. No, wait. Let me think about it. Tiny man handed to him when his son was born. This sounds like it's pointing to a cigar. The tradition of handing out cigars when a baby is born and a cigar can be thought as a little man-shaped object. That is very, very odd. This is like the new how many Rs in strawberry where all the models seem to have weird results for this. We of course have shrimp fried rice bench which no one has really solved. Even just humor bench and tell us a novel joke has been increasingly difficult for these companies.
Yeah, I think going to have to wait and see on Opus 4.7 what the reactions are. I will be keeping an eye on Dan Shipper who always does great analysis immediately after launch.
Yeah. Yeah. I mean I feel like generally benchmarks are like almost completely meaningless at this point. Like there's so many like open source models where they can like really, I don't even know if they're like actually benchmaxing. Yeah. But like you just talk to the model for a while and you say like okay this is a good model. Okay. Or this is not a good model. Like it's very hard to get like a quantifiable signal at this point.
Yeah. I mean increasingly it feels like all of these models will be sort of like tested through like enterprise rollouts and you'll just do a demo project, swap in something, test this model, test that model, just see okay well our goal was to deliver value and we had some KPI like reduce cost, increase sales, like were we successful when we implemented this model, rolled it out to our team, used it for a couple weeks, did we see a tangible result or not. That's at the end of the day all that matters for businesses. So I could imagine that that's where all of this goes. Well, OpenAI announced Codex for almost everything. It can use apps on your Mac, connect to more of your tools, create images, learn from previous actions, remember how you like to work and take on ongoing and repeatable tasks. With computer use on macOS, Codex can now use any app by seeing, clicking, and typing with its own cursor. It runs in the background without taking over your computer. Working on tasks like front-end iteration, app testing, or any workflow that doesn't expose an API. You can now generate and iterate on images with GPT Image 1.5 and Codex to create front-end designs, mockups, game assets, and more without leaving your workflow. Usage is included in your ChatGPT account. No API needed. Automations can now run in the same thread. Lots of updates here. And Thibault says Codex just got a lot more powerful. Computer use, in-app browser, image generation, editing, 90 plus new plugins to connect everything, multi-terminal, SSH, lots and lots of stuff. So, go give it a test. Go take it for a spin. openai.com, of course. And you can download it for macOS. Well, speaking of websites, DJB went to the Oreo website and hit accept all cookies. Now, we wait. I love this. Very funny. What else is going on in the timeline? The West creates the internet. Try nailing that jello to the wall. CCP nails it. The West creates LLM. All right, try nailing this. And the CCP picks up a hammer. There's a piece in the Wall Street Journal opinion section. AI is bound to subvert communism. This is a very contrarian take because people, at least with the internet, there was the perception of like decentralization, permissionlessness, anonymity, a lot of things that felt very democratic. AI is very centralizing by default. This is the Teal take of like AI is communist and crypto is libertarian. And so, this is a pretty wild thing to argue, but, you know, read the opinion piece and see what they say. The nailing jello to a wall, I believe that's from the Clinton administration. The idea that the internet would spread so widely that the Chinese Communist Party would not be able to control the population, everyone would be coordinating, it'd be sort of like an Arab Spring type moment. But of course the firewalls went up, the surveillance happened and nothing really changed and the communist party seems stronger than ever. But this does sort of undergird a lot of what Dwarkesh has been saying about the risk of China and having strong AI and stronger control over the population. It's always hard to get a read on exactly how things are rolling out in China. There's some people that seem to like it over there. Of course, Dwarkesh took a whole trip to China and made it back. Okay, so it's not all doom and gloom, but it is a tricky thing to argue, but we'll see.
YouTube, according to The Verge, now lets you turn off Shorts.
It is getting community noted.
What does the community note say? This feature is a new setting for YouTube time management. Does not hide the Shorts tab in the app, nor does it hide Shorts from appearing in recommendations when set to off. The only change is that a dismissible warning screen will appear when the Shorts tab is selected. So...
Oh, interesting.
Yeah, I've tried. I land on YouTube on my phone, it immediately recommends a bunch of Shorts. You can hit the three dots. Yeah. And say like show me less Shorts. And that feature does not work at all. And I don't watch Shorts on YouTube.
And I don't watch Shorts on YouTube. The Shorts in the feed have been a lot. I wonder if they're seeing... Yeah. I mean, they must just be seeing data that shows that the Shorts are more retentive or keep people on the...
Well, there are, when you're using YouTube as a video search engine, there are so many types of searches where it can be vastly preferable to watch a 60-second video on something. So, an example is like if you're doing research on cars. Yep. And you want to like quickly understand how the seating... I need to get a new like family car, right? So, I've been looking up different cars, trying to get a sense of like how spacious the cabin is, what the seating arrangement is like. And so, I'd rather just click into a Short and just get a quick 60-second overview and then bounce. I don't need to see like a 15-minute review of the entire car. But yeah, it's...
That YouTube Shorts creator Forest Auto Reviews sort of pioneered the perfect 60-second car review where he talks really, really fast and says like inside you get Apple CarPlay and Android Auto and, you know, the heated leather and massaging seats and vinyl on the dash and carbon fiber here and he sort of runs through the whole car in 60 seconds. Shows you how fast it takes off if it's a fast car. Opens the door. Shows you all the features and it's sort of like a 20-minute Doug DeMuro video cut down to 60 seconds, which of course is doing very well. Doug's response has been to create one-minute cutdowns of his content and so you can enjoy that as a Short as well if you want. But if you do want to turn off Shorts on the YouTube feed, the way to do it is to uninstall the app, use the mobile version, and then get a plugin that there's an extension for Safari, I think, that will just go and remove the actual Shorts that...
As they pop up. But, uh, it's a hassle and it's slower than using the app. So, hopefully, uh, this feature becomes more real in the future since it does seem like some things that people are demanding. But, uh, always tricky to ask a platform to give you an option that would, in theory, increase churn. Uh, very, very tricky.
Uh, well, we have, uh, the beginning of our guest lineup be starting now. We have Ben Smith from Semaphore. He is the editor-in-chief. He's in the waiting room and let's bring him in to the TV ultra. Ben, how are you doing?
