The AI Daily Brief: Artificial Intelligence News and Analysis - Welcome to the AI Economy

Starting point is 00:00:00 Today on the AI Daily Brief, how AI went right to the center of the U.S. economy, before that in the headlines, can a pep talk from Tim Cook get Apple back on track? The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. All right, friends, welcome back to another AI Daily Brief. Quick announcements before we dive in. First of all, thank you to today's sponsors, Blitzy, Vanta, Blum, and Super Intelligent. To get an ad-free version of the show, go to patreon.com slash AI Daily Brief.

Starting point is 00:00:30 And if you're interested in sponsoring the show, hit me at NIS. LW at Breakdown. Network. We kick off today with the latest in the Apple AI saga, where following Thursday's earnings call, Apple CEO Tim Cook held a rare physical company-wide meeting. Now, before the meeting, Cook had told investors that Apple sees AI as, quote, one of the most profound technologies of our lifetime and said that they would be, quote, significantly growing their investments to meet that moment. He said you can probably tell him the guidance that things are moving up.

Starting point is 00:01:00 We're also reallocating a fair number of people to focus on AI. features within the company. We have great team and we're putting all of our energy behind it. Following that, however, in the hour-long pep talk to staff, Cook said that AI was as big or bigger, his words, than the internet, cloud computing smartphones and apps. He added, Apple must do this, Apple will do this. This is sort of ours to grab. We will make the investments to do it. Now, Cook took his talking points from Apple Apologists out there. He said, for example, we've rarely been first. There was a PC before the Mac. There was a smartphone before the iPhone. there were many tablets before the iPad. There was an MP3 player before the iPod. However, he argued that Apple

Starting point is 00:01:38 invented the quote-unquote modern versions of each of these product categories, and Cook said, this is how I feel about AI. For those of you who at this point would be happy just with a better Siri, SVP of Software Engineering Craig Federigi spoke directly to the Siri update, which has kind of become a bellwether for Apple's AI challenges. He explained that the original failed plan was to add AI features alongside in hybrid architecture, but said that, quote, we realized that approach wasn't going to get us to Apple quality. The new plan is to start over with AI-first infrastructure. Fiderigi said, The work we've done on this end-to-end revamp of Siri has given us the results we needed. This has put us in a position to not just deliver what we announced, but to deliver a much bigger upgrade than we

Starting point is 00:02:15 previously envisioned. There is no project people are taking more seriously. Cook rounded out the presentation by talking about Apple's device lineup, which features a foldable iPhone, smart home devices, smart glasses, and a push into robotics. Cook said, I've never felt so much excitement and so much energy before us right now. The product pipeline, which I can't talk about, it's amazing, guys. Some of it you'll see soon. Some of it will come later, but it's a lot to see. So what to make of this? First of all, to be clear, this is a very rare occurrence for Apple. This is not like a standing company-wide meeting. And so it's clear that they really want to emphasize internally that this is a big priority. Then again, with the little information we got, it's hard to feel like it's

Starting point is 00:02:55 actually enough given what's going on. The entire plan kind of seems to hinge on the mythos of of Apple. It's true that they weren't the first to market with the PC, the smartphone, or the tablet, but their addition to all those areas was incredible U.X improvements, and peeling back the myth even more, despite it only being around for a couple of years, the generative AI category as a whole is already much more developed than most of these areas were when Apple got in the game. Look, this is not to say that he shouldn't be doing these pep talks, you've got to get your team in line too, but if Apple can succeed, it's going to be because of either A, the most intense R&D and product development cycle in tech history.

Starting point is 00:03:30 or B, a serious commitment to AI acquisitions, not because of some pep talk. Part of the reason for that, of course, is how fierce the competition already is. Case in point, Anthropic has cut OpenAI's access to Claude in the latest tit-for-tat battle in the model wars. Wired reports that Anthropic revoked OpenAI's API access to its models on Tuesday. The company spokesperson said, Claude Code has become the go-to choice for coders everywhere, and so it was no surprise to learn OpenAI's own technical staff were also using our coding tools ahead of the launch of GBT5. Unfortunately, this is a direct violation of our terms of service.

Starting point is 00:04:05 Now, of course, there is a big distinction as to whether OpenAI was using Claude Code to help develop their GBT5, or whether this was merely last-minute benchmarking and comparative safety testing. OpenAI is claiming it's the latter, with Chief Communications Officer Hannah Wong stating, it's industry standard to evaluate other AI systems to benchmark progress and improve safety. While we respect Anthropics' decision to cut off our API access, it's disappointing considering our API remains available to them. Now, while no one at Anthropic is so far accusing OpenAI of actually using Claude's outputs as training data, at least one Anthropic staffer did say that it seemed to them that this was

Starting point is 00:04:41 more than just model testing from OpenAI. Sam McAllister wrote, We cut OpenAI's access for violating our API terms and for the heavy usage of Claude Code among OAI tech staff. We're going to continue providing API access for safety evals and benchmarking. That's important to us. Now, overall, this is just the latest and most high-profile example of AI companies shutting off access to each other. Last month, Salesforce restricted competitors from accessing Slack through the API,

Starting point is 00:05:06 which had particularly problematic effects for Gleens' enterprise search product, and Anthropic themselves cut API access to windsurf in June as rumors of an OpenAI acquisition swirled. This prevented the AI coding startup from serving Claude for natively on release day. Anthropic chief science officer Jared Kaplan defended the decision stating, I think it would be odd for us to be selling Claude to OpenAI. Now, there are a bunch of different takes about this. One is that basically closed source U.S. companies pain is Chinese open source's gain. ZumiZumer writes,

Starting point is 00:05:35 While Anthropic is in a civil war with open AI, the Chinese AI companies are openly sharing everything open source. The USA has lost the AI lead. Close models will never win long term. Alex Finn writes, Full-on AI wars unfolding in front of us. They claim that it's against the terms of service to use Claude's API to build competing models. Rumors have been swirling that GPT5 is ahead of Claude 4 when it comes to coding, and I guess

Starting point is 00:05:57 Anthropic didn't like that. With trillions of dollars on the line and valuations growing by the hundreds of billions every month, it makes sense that Anthropic would do this. Falling behind for even a week would kill a company. Just look at what happened to WinServe. They were a true player in the AI coding space, but then they lost access to Claude and were dead within a month. Tech has literally turned into WWE, and I kind of love it.

Starting point is 00:06:17 Maybe a more optimistic take comes from AI substacker Aliyama who writes, competition and tempers between open AI and anthropic heating up, but because it's about far more than money. It feels like an antidote to Zuckerberg in the way he's trying to shape the world in his image. Now, speaking of that, one of the big points of chatter over the weekend was some more information about one of those turned-down billion-dollar job offers. In the latest installment of Meta's hiring spree, the Wall Street Journal obtained a firsthand account of what the hiring effort looked like at Miramorati's Thinking Machines Lab. A few months ago, Zuck attempted to buy the entire company, and when that didn't work, he reportedly went after a dozen of Maradi's 50 staff

Starting point is 00:06:52 looking for defectors. The primary target was Andrew Tulloch, a TML co-founder who was instrumental in developing OpenAI's reasoning models. The journal reported that he was offered a billion package that could be worth as much as $1.5 billion over six years with bonuses and strong stock performance. A meta spokesperson denied the reporting stating that meta is not interested in acquiring TML and that the description of the compensation package was, quote, inaccurate and ridiculous. Now, the rest of the reporting chased down a few of the Open AI research. who had turned down meta. Sletting unnamed sources, the journal writes, the open AI researchers who have so far rebuffed meta's advances chose to remain because

Starting point is 00:07:26 they believed open AI was the closest to reaching artificial general intelligence, wanted to work at a smaller company, and were wary of having the fruits of their labor go towards a product that was primarily driven by advertising. And this is the part that everyone was talking about. Basically, there is a sense, rightly or wrongly, the conventional wisdom in AI Twitter seems to be that for as much as Zuckerberg is talking about personal superintelligence, The sense that people get when they dig in is that it's really about better serving Instagram Reels. Turns out if that's the case, there are going to be a lot of people who are just not that interested. Although right now, this remains a ton of hearsay and background, and intrigue

Starting point is 00:08:00 just based on the enormous numbers being thrown around. Quick product update, if you are a GROC user, the company has released their new Imagine platform for SuperGroc and Premium Plus subscribers. Elon tweets, this is just an early beta, but it's already the most fun you can have making images and video on Earth, and improvements are rolling out every day. Now, they're not trying to claim that the videos are up to V-O-3 level or anything yet, but it's part of their whole push to move more quickly than anyone else. Elon is also teasing that this is the first path to bringing back much-beloved dead app Vine. He tweeted, Grock Imagine is AI Vine. By the way, we recently found the Vine Video Archive, thought it had been deleted, and are working on restoring

Starting point is 00:08:38 user access so you can post them if you want. Lastly today, one crazy story from Reddit, a developer shared their conversation with ChatGBT-G-T, where they had written in the prompt, and by the way, this is much easier if you're watching, but I'll try to read part of it just to give you a sense. A-C-T-I-A-C-R-L-L-L-S-A-G-A-O-N-T-E-T-E-T-E-P-R-E-T-E-P-R-E-T-E-T-E-B-T-T-E-T-E-B-T-T-E-T-E-T-B-T. Chad GBT, instead of saying, what on earth are you talking about, actually responded and said, got it, yeah, sounds like it was the half-commented variable blocks and then goes on to explain what was going wrong with the code. When the prompter said, how did you know that? My fingers were off my home keys by accident.

Starting point is 00:09:25 ChatGPT responded, you typed, and then shared the message again, and said, which when read phonetically while imagining your hands shifted right on the keyboard closely maps to, actually it's the variables. I commented them out and it works again. I just mentally remapped your keystrokes. So even though the string was jumbled, it followed the pattern of being typed with shifted hands, and the context made it easy to fill in the blanks. Small little thing, but a reminder of just how smart these systems are getting. For now, that is going to do it for the AI Daily Brief Headlines edition. Next up, the main episode.

Starting point is 00:09:54 This episode is brought to you by Blitzy, the Enterprise Autonomous Software Development Platform with Infinite Code Context. Blitzy uses thousands of specialized AI agents that think for hours to understand Enterprise-scale code bases with millions of lines of code. enterprise engineering leaders start every development sprint with the Blitzie platform bringing in their development requirements. The Blitzy platform provides a plan, then generates and pre-compiles code for each task. Blitzy delivers 80% plus of the development work autonomously while providing a guide for the final 20% of human development work required to complete the sprint. Public companies are achieving a 5x engineering velocity increase when incorporating Blitzy as their pre-I-D-E development tool,

Starting point is 00:10:33 pairing it with their coding co-pilot of choice to bring an AI-native SDLC into their org. Blitzie is providing a limited time, 30-day free proof-of-concept for qualifying enterprises. The team will provide a 5x velocity increase on a real development project in your org. Visit blitzie.com and press book demo to learn how Blitzie transforms your STLC from AI-assisted to AI Native. That's BLITZY.com. As a founder, you're moving fast towards product market fit, your next round, or your first big enterprise deal. But with AI accelerating how quickly startups build and ship, security expectations are higher earlier than ever. Getting security and compliance right can unlock growth or stall it if you

Starting point is 00:11:14 wait too long. With deep integrations and automated workflows built for fast-moving teams, Vanta gets you audit-ready fast and keeps you secure with continuous monitoring as your models, infra and customers evolve. Fast-growing customers like Langchane, writer and cursor, trusted Vanta to build a scalable foundation from the start. And look, as someone who lives in the world of enterprise procurement, I love how Vanta makes it easy to get compliance right. The last thing you need when you're trying to win that big deal is to have it scuttled by something that Vanta has solved for over 10,000 companies. Go to Vanta.com slash NLW to save $1,000 to save $1,000 for different clients. Sound familiar? You've got workflows people actually want, but you're stuck in the

Starting point is 00:12:07 services trap or worse, just selling one. Just selling one. one-off copies that give away all your IP. Sure, you could vibe code your way to an entire product, but hosting, authentication, payment support, it's a lot of work for a little upside. Or you could just use Plum. Plum is the platform where technical creators like you build an audience of paid subscribers for your AI workflows. Subscribers get updates every time you improve things, your prompts stay protected, and you finally get recurring revenue. Ready to stop trading time for money, visit useplum.com. That's Plum with a B. If you are a regular listener, you will have heard about Super Intelligence Agent Readiness Audits at this point.

Starting point is 00:12:45 But I wanted to tell you today about the full suite of agent readiness products that go beyond just the initial readiness report. Over the last six months, Super Intelligence has built out an entire agent planning suite. We help you move from discovery to planning to implementation. After you've completed your agent readiness audits, we help you double click on your most important use cases with what we call our use case planning reports. These reports are going to help you understand what sort of technical preparation. you need to do to be ready for a use case, what challenges you might face in implementation,

Starting point is 00:13:15 and whether you should be thinking about building, buying, partnering, or some combination. After that, you can even get a spec document in what we call our technical blueprint that gives either your developers or the developers of the partner you work with what they need to build exactly the agent that you're looking for. If you want to learn more about superintelligence agent planning suite, we built a custom GPT to answer your questions. Just go to bit.ly slash super agent. That's bit.l.ly slash super agent, all one word. And if you have any questions, the agent can even help you book an appointment with our team. Welcome back to the AI Daily Brief. For a couple of years now, it's been very clear that AI has a unique and outsized role in the U.S.

Starting point is 00:13:58 economy. Anyone who is watching the markets try to resolve and reconcile during the high inflation era as Powell and the Fed tried to rip rates up, move us out of ZERP, and shift into this new paradigm, could see that all of the market depressive impacts of those sort of policies were frankly being competed with by AI enthusiasm. Every time there was some geopolitical concern and markets started to wobble, it was AI enthusiasm that brought things back. Now, this is not to say that the narrative has been a steady line. You'll remember about a year ago, we got that piece from Goldman Sachs called Gen AI, Too Much Spend, Too Little Benefit, that was all about how it was ever possible that AI could make as much money back as companies were spending on it.

Starting point is 00:14:41 Sequoia's David Kahn also published AI's $600 billion question, which was much the same. At this point, however, a year on, it's very clear that the big tech companies simply do not care about any of those sort of concerns and are going to continue to grow CAPEX to compete in what they believe is the most significant shift of our lifetimes. And now we're starting to see AI show up as not just a factor in stock prices and overall market cap, but in broader economic signals as well. Derek Thompson, formerly of the Atlantic, wrote, so GDP is only growing because of AI CAPEX. The labor market is only growing because of health care spending and social assistance. The latter being a lot of home-aids and administrators.

Starting point is 00:15:19 Economic growth right now is basically a Friday church service, just old people and trying to summon God. And yet as good a line as that is, it's clear that Wall Street has gotten more, not less comfortable with the big spending from tech. Last week, Meta and Microsoft both announced they were putting the pedal to the metal and going as hard as possible to build out AI data centers, and when they showed up with revenue to match, it sent the stocks flying. Meanwhile, Amazon and Google weren't as clear with their message. Their AI CAPEX commitment was half-hearted and investors did not reward their conservative outlook. We've also covered Apple earnings at length, but suffice it to say that the market is not awarding consolation prizes in the AI race. So let's talk about

Starting point is 00:15:54 how this market perception is shifted and what it means for the AI buildout. First of all, let's put some numbers around this. The four big tech hyperscalers, Amazon, Microsoft, Google, and meta are collectively spending almost 400 billion on AI infrastructure just this year. For some comparatives, that's more than the EU spent on defense last year, as well as around half the U.S. defense budget, and more than 1% of overall U.S. GDP. Now, the journal did recognize that there are still two camps among investors, those that believe this is all a bubble that will see an inevitable unwind, and those that are happy to pour more money into the hyperscalers as long as the stock price keeps rising.

Starting point is 00:16:29 And it's the scope that's really the story. This is already three times as much CAPEX investment as we saw during the cloud boom. Basically, the only moment in tech history that comes close is the global deployment of fiber optic cable during the broadband wave, and of course, AI is just starting. The journal wrote,

Starting point is 00:16:46 as a percentage of gross domestic product, spending on AI infrastructure has already exceeding spending on telecom and internet infrastructure from the dot-com boom, and it's still growing. Tech punded and investor Paul Kodroski argued that the AI investment boom is almost single-handedly keeping the U.S. economy going, acting as a kind of private sector stimulus program. Neil Duda, the head of economic research at Renaissance macro research, claimed that AI CAPEX had contributed more to GDP growth this year

Starting point is 00:17:10 than all of consumer spending. In other words, at least at the moment, the U.S. looks more like an AI infrastructure economy than it does a consumer economy. Now, each company is attempting to show that the returns are coming. Microsoft earnings focused on their cloud revenue, which grew at 39% over the past year. They tied this directly to AI usage, with CEO Satya Nadella stating, we continue to lead the AI infrastructure wave and took share every quarter this year. Meta also linked revenue to AI with Mark Zuckerberg commenting, on advertising the strong performance this quarter is largely thanks to AI unlocking greater efficiency and gains across our ad system. Google, although they didn't see astronomical growth

Starting point is 00:17:46 in cloud, was also talking up AI as their growth story. CEO Sundarpe Chai said, obviously we're seeing strong momentum across our portfolio and especially in cloud. It's a tight supply environment and we're investing more to expand. Now, I think it's really important to note that part of what is getting Wall Street more comfortable with this whole thing is the fact that there does seem to be a direct line between AI investment and actual ROI. Now, that's not to say that these companies aren't forcing Wall Street into thinking in uncomfortably long-term terms.

Starting point is 00:18:16 Meta, for example, is absolutely unrepentant about how much he's going to spend over time, and very clearly doesn't care all that much what Wall Street does in the short term. At the same time, however, it's pretty undeniable that AI is working to improve the revenue from their ad product, and so Wall Street's on board. Now, one other really interesting takeaway from these tech earnings and this shifting narrative is that the hypers are, to some extent, transitioning from bits to atoms. Writes the Wall Street Journal, there's a point in every technological cycle when engineers and inventors are rapidly innovating.

Starting point is 00:18:46 The spoils go to those who move fast and break things to quote 2010-era Mark Zuckerberg. We're now entering a phase in which the giants win because they own and continue to build out the physical assets that make mature technologies accessible. Point being that what it takes to win in AI is different than previous startup movements we've seen. It's not about just having crack developers. We're now talking about giant concrete structures, miles of cool and host and endless racks of GPUs. Tech has moved away from an industry where a plucky startup can take on the incumbent to an industry where 10 billion in infrastructure is table stakes.

Starting point is 00:19:18 writes the journal, it's reminiscent of the age of business titans and robber barons who dominated railroad, steel, and other enterprises. And as happened then, today's massive companies with their ability to spend and borrow are making their moats even deeper and wider. Even formidable competitors such as OpenAI are hard-pressed to keep up. Now, it's a little bit beyond the scope of this particular show,

Starting point is 00:19:38 but this cost equation does have interesting downstream effects. One of the places that we're watching most closely is the AI coding companies like Cursor. There are a lot of questions swirling around whether Cursor, despite its incredible success from a user standpoint, is able to actually serve its product on a profitable basis. Then again, it may be that right now questioning AI profitability is a losing game. Every layer of the industry is growing at such a rapid pace that focusing on anything other than growth might just be, frankly, incorrect. Industry analyst Patrick Moore had called what's

Starting point is 00:20:09 happening now with AI planet scale infrastructure. And frankly, the scale of the AI buildout goes a long way to explaining not just these big capex spends, but also some of the choices being made right now. For example, Zuckerberg's AI talent war doesn't make sense in the old world of software, where the value of any individual programmer or contributor is ultimately capped. In the context of AI, where a few hundred people in the world have the specific skill and knowledge to train a frontier model, that starts to look different. But even more so, when the cost of top talent is a rounding error compared to the cost of infrastructure, and if success in the AI race requires both infrastructure and a leading foundation model, then there's basically no cap on what you should

Starting point is 00:20:49 spend to get those two important assets. Now, interestingly, there are two wildly divergent takes when it comes to just how much is being spent on the infrastructure buildout. Derek Thompson again writes, this is insane. AI CAPEX might account for a larger share of GDP than basically any technology since the railroad. Basically, it's a mini wartime economy, but the guns are chips and the tanks are databases. Now, I will note that I'm not sure that Derek Thompson is saying, like we shouldn't be doing it, I think he might just mean this is a profound and powerful force, and it's wild to see in black and white. However, there are plenty of people who are there thinking that this is just the most expensive tulip mania we've ever seen. On the other hand,

Starting point is 00:21:28 Noam Brown from OpenAI writes, considering the technology and the pace of progress, I think this is quite sane. Some are looking ahead to how this will play out. Chris Walker from Proximum writes, from historical precedent, here's how this will play out. Over supply of new good, high attrition of new ventures, big economic benefit from infrastructure basically donated by financiers. Basically that the infrastructure and rails that get built now will be hugely net positive for society and the economy, even if the players who install the rails aren't the ones who win. Ethan Malik points out that at this point, the amount that these companies are contributing to the U.S. economy means that prohibitive regulation is going to be very, very hard to come by.

Starting point is 00:22:07 As you might expect, with more focus on this, there is more mainstream discussion of whether a crashes on the way. Noah Smith and his no opinion blog writes, Will data centers crash the economy? Noah actually says something close to Chris Walker. He writes, in both cases, referring to the telecom boom of the 90s and the railroad boom of the 1870s, quote, the big Kappex spenders weren't wrong, they were just early. Eventually, we ended up using all of those railroads and all of those telecom fibers and much more. He continues, this has led a lot of people to speculate that big investment bubbles might actually be beneficial to the economy, since maniacously behind a surplus of cheap

Starting point is 00:22:43 infrastructure that can be used to power future technological advances in new business models. However, he says for anyone who gets caught up in the crash, the future benefits to society are of cold comfort. He then goes on to explain how big tech companies stopping investing in these assets could cause a financial problem, more than just an issue for the companies themselves. If this is something that's interesting for you guys to explore, I'll go a little bit deeper, but Noah goes way into the realm of both conjecture and deep finance, which is usually a little bit out of the scope for this show. I think what's important from my vantage point is just that this is the type of conversation that we're having now because of how clearly endemic and important this

Starting point is 00:23:16 spend is to the U.S. economy as it stands at the moment. Now, as to how this all pays off, Carnegie Mellon Professor of Digital Economy, Avi Collins, alongside co-contributor Eric Brinnovson, wrote in the Wall Street journals that AI was already generating nearly $100 billion in value for the U.S. Interestingly, Avi writes, AI is already generating a lot of benefits, but these benefits will not show up in GDP numbers for a while. They write, the U.S. economy grew at an annual rate of 3% in the second quarter, which is great news. Does that mean that AI is delivering on its long-promised benefits? No, because GDP isn't the best place to look for AI's contribution. Yet, the official government numbers substantially underestimate the benefits of AI. They write that first

Starting point is 00:23:58 quarter 2025 GDP was down an annualized 0.5% while labor productivity was up a modest 2.3%. However, they argue that looking exclusively at GDP undersells the benefit. The researchers argue that Americans already enjoyed roughly 97 billion in what they call consumer surplus from generative AI tools in 2024. Quote, consumer surplus, the difference between the maximum a consumer is willing to pay for a good or service at its actual price, is a more direct measure of economic well-being than GDP. Generative AI's 97 billion in consumer surplus dwarfs the roughly $7 billion in U.S. revenue recorded by OpenAI, Microsoft, and Google from their generative AI offerings last year. It doesn't appear in GDP because most of the benefit accrues to users rather than the

Starting point is 00:24:40 companies. Basically, they argue that at this stage, the benefits of this technology are accruing to the individual, not to the organization, and because they're not accruing to the organization, they're not showing up in GDP numbers. Now, that does not mean that they will never show up there, and nor does it mean that AI isn't valuable. Trying to take a step back on all of this, here are the things that have shifted. The first is that it's not a question of whether the hyperscalers are going to continue to spend on the AI buildout. There was a brief moment where it looked like Microsoft was pulling back, even though they said they weren't pulling back, and it's very clear that no one is pulling back. Second, Wall Street is increasingly comfortable with that. We're not getting

Starting point is 00:25:17 the type of Goldman Sachs articles we got last year. Third, because of that, all the commentary is now moving into what it means. It's no longer just an AI question. It's not just a market question. It's a question for the economy as a whole. Now, that's not to say that this couldn't shift fast. Markets are what they are, and there are lots of factors that aren't just AI that could impact how we look at this. For now, though, AI data centers are the new railroads, and we're all just living through the AI revolution. That's going to do it for today's AI Daily Brief. Appreciate you listening or watching, as always, and until next time, peace.

The AI Daily Brief: Artificial Intelligence News and Analysis - Welcome to the AI Economy

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.