How the Latest AI Leap Is Disrupting Business Models

Back at the end of 2025, we got a set of model advances which first seemed innocuously like just the next step of an incremental gain, but which, as we would discover a few weeks later when people went home for the holidays and really started to experiment, represented something totally new in terms of the capability set. Those model jumps would lead to the agentic progress that has defined 2026. And now it seems we might be in a similar moment. More than ever before, people are having success decompiling software and particularly games and reconstituting them in new ways using models like Astra and Opus 5.5. Today we're exploring what the implications are for the business model of these software and game companies and how they're responding to the new challenge. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. All right, friends, quick announcements before we dive in. First of all, thank you to today's sponsors, KPMG, Harbor, Granola, and Blitzy. To get an ad-free version of the show, go to patreon.com/AIDailyBrief, or you can subscribe on Apple Podcasts. And to learn more about sponsoring the show, send us a note at sponsors@aidailybrief.ai. We kick off today with a category of story that is going to be frequent for some time, which is of course the latest pre-IPO revenue numbers from Anthropic and all of the discourse that surrounds it. On Wednesday, Anthropic hosted institutional money managers at their San Francisco office for a pre-IPO investor day. The event served as an opportunity to meet the executive team and learn more about the company. They reportedly handed out thinking caps and managed to avoid talking about existential risks, but all eyes were on the latest revenue figures delivered at a presentation towards the end of the day. The Financial Times reports that Anthropic told the investors they'd recorded $16.8 billion in revenue in Q3, up 45% from their Q2 number. That seems to be the actual revenue generated, not an annualized run rate. However, presumably it's still gross revenue counted before the revenue split with Amazon and other compute partners. as that is Anthropic's preferred accounting model. Anthropic also said that they were profitable on an adjusted basis, are generating positive free cash flow, and had 85% adjusted gross margins and 11% adjusted operating margins. Anthropic has typically excluded stock comp, training costs, and revenue shares from their adjusted numbers. In a little preview of how things are going to be when these companies go public, the report was treated like the most important company in the world was delivering their quarterly earnings report. And in many ways, that's not too far from the truth. The stock market has been shaky throughout the summer and early fall, so a lackluster IPO from Anthropic could have some serious consequences. Still, it is certainly unprecedented to see so much focus on revenue rumors from a private company. Some analysts on X were quick to call it a big flop. With tech stocks selling off after hours, Geiger Capital posted, not good enough. It's over. Others wondered how in the world a 50% quarter-over-quarter growth rate was perceived as an underperformance. Abhi Desai of Krypton Research commented, market doesn't like $17 billion in quarterly revenue from Anthropic at this stage? Seriously? Adding a further wrinkle, after the initial number, Reuters added that Anthropic is expecting $110 billion in annualized revenue by the end of the year. Their Q3 numbers extrapolate out to $70 billion annualized, implying that Anthropic believes they have another 57% growth spurt coming in Q4. Ticker Trends updated their live revenue chart, which has been pretty accurate over the past 6 months. It showed that revenue growth has been flat since the end of July, meaning Anthropic must see something big coming that will juice revenues into the end of the year. Andrew Curran thought it was pretty obvious what that would be, posting, $110 billion is back on the menu. They must have a lot of confidence in their next model release, the wizard, Fable 5.5. As I said before, it is unusual in the extreme for Wall Street to be pinning their hopes on unaudited financials from a single private company, but That's how the market is trading right now. Anthropic is still planning on going public at a $2 trillion valuation sometime over the coming months and will reportedly ask for record-breaking IPO funding. It's worth noting that there is an accounting benefit to getting the deal done quickly. Anthropic can go public with a financial disclosure up to 134 days old, so there's a strong chance they'll only include Q1 and Q2 numbers. That means the unaudited Q3 numbers reported on Wednesday will be the most recent information the market has. Of course, none of that matters if Fable 5.5 is a hit and Anthropic can leak booming new revenue as a result, but right now is their window of opportunity to get the IPO done at top dollar. Continuing on, a lot of the big headlines from yesterday were in the market sphere. CoreWeave's bonds have reached distressed levels as NeoCloud funding dries up. On Tuesday, unsecured CoreWeave bonds maturing in 2032 plunged to $0.80 on the dollar in a rapid move. They're now beyond junk debt, trading at an implied interest rate of 15%, indicating the market is pricing a serious risk of default. Investor Chamath Palihapitiya posted a chart of the past 4 months of price action, showing a rapid decline from 10% rates. Elon Musk simply responded, yikes. The stock was also down 15% on the day, outpacing a 1.6% fall overall in the Nasdaq. AI bears are piling on the declining Neo Cloud. DA Davidson analyst Gil Luria, who has referred to CoreWeave as the subprime tranche of the AI industry, warned the stock could go to zero if data center funding stalls. He argued that the company is destroying value at every turn, his phrase, by continuing to borrow, adding, If CoreWeave is borrowing at 9% and the best return they can get off that is a 5% return, it doesn't matter how much they scale. Borrowing at very high interest rates in order to build data centers that are increasingly commoditized outside of the hyperscalers may not end up being a good investment. The more leverage we use to build data centers, the more risk we're bringing into the system. CoreWeave CEO Mike Entrater is fighting the market, claiming it's acting irrationally. In a staff memo shared with Bloomberg, Entrater wrote, when you're building a new type of technology company, there will be periods of time when the market misunderstands. We are confident in our business model, our customers, the extraordinary quality of our execution, and the relentless discipline around the financial engine we have built. The Information, however, noted that Entrater's actions don't line up with his defiant words. According to SEC filings, Entrater has sold CoreWeave stock every week since April, And other insiders have been cashing out as well. CoreWeave previously said in a statement that the company's co-founders were, quote, making personal financial planning decisions that any responsible founder with concentrated holdings would make. And while that may be true, it is certainly not helping the stock pull out of this downturn. On the flip side, hold up the market dynamics. There is certainly no decrease in the demand for AI. In fact, Dutch chip fab manufacturer ASML says that the company won't be able to meet demand for several years. ASML is the leading producer of the complex machinery behind semiconductor manufacturing, referred to as tools. For high-end technology like the EUV lithography required for AI GPUs production, they hold a complete monopoly. ASML had previously delivered a plan to boost supply by 30%. However, during this week's earnings call, they warned that demand is running way ahead of what they can manage. TSMC, Samsung, and SK Hynix have all laid out ambitious expansion plans over the coming years. Then there's Elon Musk's TerraFab project, which aims to be the largest chipmaking facility in the world. ASML has said that they won't be able to meet this demand for at least 2 years, if not longer. CEO Christophe Fouquet said, Customers are asking us every single day to get more tools. We are investigating ways to increase capacity for 2028 beyond the 30% increase we previously communicated. He added that some customers have reservations that stretch into 2029 and even 2039. CFO Roger Dassen noted that there's nothing much that can be done to bring on extra supply in the capital-intensive industry, commenting that it would be, quote, Extremely hard to get something done that gives you tremendous capacity by 2028, even if we started digging today. In the latest on AI safety discourse, Nvidia CEO Jensen Huang is doubling down on the idea that it is the responsibility of the companies that are producing the AI to make sure their products are safe. During a string of appearances in the UK on Wednesday, the Nvidia CEO was repeatedly asked his views on AI safety. In a sit-down interview with Bloomberg, Jensen was asked if he would stop supplying chips to an irresponsible AI developer. It's a hypothetical, but yes. But before you do so, you would have a conversation. We don't run other people's companies, and there's a limit to the judgment we apply to other companies' use of our technology. Our technology is general purpose. During a press scrum later in the day, a journalist asked Jensen for more details. The journalist summarized his understanding of Jensen's view as, there's a 0% chance of extinction, AI safety is just an engineering problem, and we've got it solved. No, what I say is that it is in our control. Between now and the end of humanity, we're going to have a lot of work to do. Every single day as we advance the technology and capability, we also have to advance the safety technology. Advancing capability and advancing safety are not in conflict. They are, in fact, one. The more nuanced discussion suggested to some that Jensen is not flippant about safety risks. He just believes that they're a tractable problem that should be engaged with engineering rather than politics. It's not as simple in his eyes as there being well-defined pro-safety and anti-safety camps in direct opposition to each other. Ill-defined though the term may be, given that I am the one making it up, I might in some ways label Jensen an AI realist. Demonstrating the point, Nvidia announced $250 million in compute for safety efforts in the UK across 5 years. Jensen said in a statement, The UK brings world-class scientific rigor to this essential engineering challenge. With this $250 million commitment, Nvidia will put accelerated computing in the hands of UK researchers to develop, test, and strengthen safeguards for increasingly powerful AI. In another example of how companies are addressing the here-and-now challenges of AI, the technology is actually causing a hiring boom in the banking sector, but not for the reasons you might think. Bank of America revealed that they've hired 1,000 contractors to support 3,000 staff in a massive new effort to patch bugs revealed by Mythos. The software patching combined with a rise in anti-money laundering compliance drove hundreds of millions of dollars in additional costs in the 3rd quarter, although this is expected to fall over time. CEO Brian Moynihan said, that risk is real risk, and that's why when the Frontier-type models came out, a lot of us sat down and said, how do we figure out how to get through this quickly? The 1,000 contractors were added in the second quarter and are all still engaged in the project today. This is reminiscent of comments earlier this month from JPMorgan CEO Jamie Dimon, who said that cyber risks in banking went up, in his words, tenfold after Mythos. He added that his bank isn't getting hysterical over the risk, But quote, what we're doing is rolling up our sleeves and going to work to fix it. One little human interest slash true crime follow-up from yesterday's story about the OpenAI employee charged with planting a listening device in their office. That accused employee had planned to skip town on the next flight to Saudi Arabia when he was arrested last week. New court documents claim that the Saudi national had booked a one-way ticket to Saudi Arabia for October 12th. The DOJ had requested a sealed indictment and arrest warrant on October 8th. Claiming that public disclosure might jeopardize the defendant's arrest. Government attorneys wrote, the defendant has the means and will have an incentive to flee the United States sooner if he learns about this case before he is arrested. It seems the instinct was correct. The arrest was carried out the following day, hitting a narrow window before the defendant could flee the country. The dramatic arrest has put international espionage front and center as a concern for the AI race, but the companies building the technology have been vigilant about foreign spies for years. A Wall Street Journal article discussed a range of different ways foreign agents have attempted to gain inside information from AI researchers. These included attractive women approaching them in bars, to spies disguised as a family at the playground. OpenAI has an entire section of their employee onboarding devoted to detecting insider threats and recognizing intelligence agents while out and about. Lastly today, bad news AI bears, the Russian government is coming for you! Russian authorities are cracking down on AI bears. Not the traders who are shorting Neo Clouds this week, but rather literal AI-generated images of bears. Reuters reports that a viral image of a bear roaming through the streets of a remote village triggered a real-life bear hunt from animal control. The image began as a prank among friends, but spread like wildfire across local group chats, prompting policy reports. Once the village was confirmed to be bear-free, the woman who confessed to creating the image was fined 15,000 rubles, around $178. And that is going to do it for the headlines. Next up, the main episode. A new study from KPMG and the University of Texas at Austin found that when people work with AI, similar skills don't guarantee similar outcomes. Researchers studied more than 500 early-career professionals and found that the best performers consistently amplified the value of AI by guiding, evaluating, and refining its outputs. These top performers, called AI Amplifiers, weren't defined by what they knew alone, but by how they worked with AI. Learn more about what separates AI Amplifiers from everyone else at kpmg.com/us/aiamplifiers. If you listen to this show, you likely have a thesis. Maybe it's enterprise adoption, maybe it's compute, maybe it's a specific lab. Harbor Capital's AI Lab Ecosystem ETFs let you express it via 5 actively managed ETFs. Each seeking exposure to the ecosystem around one major lab: Anthropic, OpenAI, DeepMind, Meta, or SpaceX AI. Your view of the AI race in ETF form. Harbor Capital Advisors' AI Lab Ecosystem ETF suite gives investors a way to invest in the AI ecosystem they believe is best positioned for success. Search Harbor AI Lab Ecosystems ETFs wherever you invest or follow @HarborCapital on X to learn more. Visit harborcapital.com for a prospectus containing investment objectives, risks, fees, expenses, and other important information. Read and consider it carefully before investing. Risks include principal loss and artificial intelligence-related risks. Harbor ETFs are distributed by Foresight Fund Services LLC. Harbor is not affiliated with AI Daily Brief, and the funds are not affiliated with, sponsored by, or endorsed by any AI lab. This is a paid advertisement and not personalized investment advice. Investing involves risk, including possible loss of principal. When I'm in a meeting, I'm fully in it. I'm thinking about the iteration and creative back and forth it takes to actually push a goal forward. What I'm not thinking about is capturing takeaways, tracking to-dos, or Any of that. And that's where Granola comes in. Granola is an AI-powered notepad that captures what happens in your meetings and turns it into clean, structured notes with the decisions and action items pulled out and easy to find. There's no setup and no configuration. It just fits into how you already work. For me, it means I get to stay in idea mode and Granola makes sure those ideas actually become action. Once you try Granola on a first meeting, it is hard to go without. You can try it totally free at granola.ai/brief. That's granola.ai/brief to get your time back. Blitzy's deep codebase understanding unlocks the thing every roadmap owner cares about: shipping new features. Here's the truth about building inside a massive enterprise codebase: writing code was never the bottleneck. Context is. Which system does this touch? Which contracts can't break? Which standards apply? Blitzy already knows because it reverse-engineered your entire codebase into a dynamic knowledge graph before feature work began. With that complete picture, Blitzy builds features end-to-end. Architecture, APIs, UI, and tests all validated against your existing systems. One Blitzy customer built an AI-native application from scratch with 100% autonomous completion, saving over 2,700 engineering hours. Features that respect your codebase instead of fighting it. Stop letting your backlog grow faster than your team. Accelerate your roadmap at blitzy.com. That's B-L-I-T-Z-Y dot com. Welcome back to the AI Daily Brief. Since pretty much the day that ChatGPT launched, people have been speculating about how AI would eventually disrupt or change or completely destroy certain types of business models. And yet, for a lot of the time since then, that conversation has been largely theoretical. There have been a few moments, however, when some big increase in capability has resurfaced that conversation in a massive way, and we are living through one of those right now. Earlier this month, an enterprising developer threatened to take down one of the largest software firms with nothing more than a Claude code subscription and a grudge against SaaS. Software is over, declared Brandon Thomas in a Hacker News post. I just one-shotted Photoshop, the whole thing. Thomas had used Claude Opus to decompile a copy of Photoshop and reconstruct it in what's known as a clean room rebuild. Prior to the AI revolution, clean room rebuilds were a seldom used way of getting around software copyright. One group of engineers would decompile software code, document features, and pass the information to another group. Having never seen the actual code, this second group could build their own version of the software without falling foul of a copyright infringement claim. As you might imagine, this was a time-consuming and costly process, so was seldom used. But AI has made it a bit more practical to achieve on an individual basis. Thomas kept working on the project and quickly had AI-generated bootlegs of the entire Adobe. There are still plenty of bugs and it's not quite feature complete, but Thomas is now offering a free version of an enterprise software suite that Adobe prices at $100 a month. This all on its own was enough to generate a mini version of the SaaS-pocalypse, with Adobe stock opening 3% lower on Monday after the story went viral over the weekend. It has recovered a little since, but there's still some fairly deep concerns about the viability of the software industry. On the All-In podcast last weekend, Chamath Palihapitiya said, I think that IP and software was effectively rendered worthless 2 days ago. Because if you think about it, when everything is headless, so when Elon makes his service headless, Benioff made Salesforce headless a few months earlier, when all of these things are headless and you can wrap them in agent tools and you can connect them via MCP to other experiences, the value of IP does no longer exist in software. It doesn't exist. And I think that probably is the most profound implication. He went on to explain that his AI SaaS business has stopped negotiating to retain IP when they build systems for clients. Before, he said, it was always a negotiation and a set of trade-offs. Now it's a who cares thing where like the IP is irrelevant. Now maybe this is a little bit, shall we say, podcast hyperbolic. People have of course been pirating Adobe software the old-fashioned way for decades. And it hasn't impacted their business. The reality is, of course, that most SaaS companies aren't selling the software, they're selling the support that goes around it— the ability for an IT procurement manager, in other words, to offload all the maintenance work for a reasonable monthly fee. Essentially, we're seeing a rerun of the vibe coding narrative, when some people briefly thought companies would be prompting their own CRM and accounting software. It's been 9 months since that narrative kicked off, and most of the companies who experimented with the idea have quietly returned to their Salesforce contracts. Which is not to say that there aren't major changes afoot. Although I'd bet on enterprise SaaS not going anywhere fast, the video games industry is facing a different set of challenges. AAA gaming is in the grips of a severe downturn coming off of the pandemic high. Console sales are at a 13-year low, major studios have laid off tens of thousands of workers, and outside of a few key titles, new games are seeing some rough sales numbers. In that context, the trend of decoupling software— has sort of hit the industry like a freight train. At first, it was some lighthearted fun with video game mashups like Super Mario in Elden Ring and Spider-Man in Arkham Asylum. Then traditional crackers used AI to accelerate efforts to jailbreak the PS5, allowing it to play games from basically any system. There are probably a few crisis meetings going on this week at Sony HQ. Analysts at Bloomberg Intelligence believe the convergence of streamed games and AI will ultimately kill off gaming consoles. The thesis is that AI will support more complex AAA games that are better served from data center hardware than a GPU in a plastic box in your living room. Consoles are also suffering from extreme cost pressure due to the competition for semiconductors. The base model PS5 is now $650 compared to $500 at launch, and that's to say nothing of the $1,000 price tag for the Steam Machine. Bloomberg Intelligence wrote, Game hardware is shifting towards a narrower market of enthusiasts. Consoles will remain relevant for avid players seeking premium performance, and perhaps among collectors in the same way that record players continue to be used for music. But consoles' role is moving from video gaming's main gateway, to just one of many access points, and a relatively uneconomic one at that. Still, business model disruption and changes in the way that things happen does not always mean for the worse. Some believe the industry will go in the other direction, with small teams becoming capable of delivering premium products. George Cal, a game engine developer, posted: Opus 5.5 is scary good at game dev. Not just code— shaders, system design, level logic, sounds, tooling, debugging, across the board. We're at an inflection point. In 1-2 years, every solo dev will be able to ship the game in their head. High quality, no studio needed. Will there be any moat left? Probably only taste and audience. Elon Musk quibbled with the timeline, commenting: Next year. Summing up the State of Play, AI news aggregator Andrew Curran wrote, if you don't use the models or watch the demos people post on social media, you may not realize there has been an astonishing jump in game development capability lately. We really are not that far away from a one-person game studio anymore. You can mash up games now and the model will deliver a stable, playable fusion in one day. Some of them are so good that people don't post them openly online because they are afraid of getting sued. That's how quickly things have improved over the last few weeks with Astra and Opus 55. Because of how virulently anti-AI the game industry has become, people are going to get completely blindsided by what's about to arrive. Not only are they unaware of how fast things are improving, they don't seem to understand the endgame. I see people talking about supply gluts, things like that. This isn't about making games to sell on Steam and compete with the industry. If the player knows the type of game they want, and their model can code the game for them at the AAA level, in the genre they ask for, with the aesthetics and character design they prefer, Then the question becomes, why buy a game from a studio at all? I think Sir Demis sees this future with Genie. For the price of your subscription, you will get any style of game you want, tailored to you. Once that starts happening at scale, most game studios will very rapidly cease to exist. And that's the real endgame. Not competition, destruction. And it's closer than people think. So how are the game studios handling this? Well, there's a few different approaches, but all of them seem to know a seismic change is approaching. Fortnite makers Epic Games have been all in on AI as a productivity enhancer for several years, but with the state of the current technology, they're making an even bigger push. This week, they banned handwritten code. Alongside this new mandate, Epic delayed the release of a forthcoming Disney game, a shooter featuring Marvel and Star Wars characters duking it out. Developers on the game have been told that they're switching to an AI-first production model, even down to the art, which will be entirely generated with full AI workflows. The reaction from the outspokenly anti-AI gaming community is perhaps what you'd expect. CEO Tim Sweeney wrote, we have a huge pivot underway here at Epic, moving developers around to retool for maximum AI acceleration. The same isn't to do the same stuff with fewer people, but to enable everyone here to build more and better stuff, and to bring the benefits to all teams using Unreal. When indie game developer Andrej Apančić responded, your inability to read the room is really, really astounding, Tim, Sweeney actually fired back. He wrote, Much of the room wants it to be 2016 again, but it's not. Our job, as the provider of tools and technology to much of the industry, is to design for the present that is and the future that will soon be. Personally, I'm optimistic about it and see it as empowering for creators. And yet, so far it hasn't been smooth sailing. This week Epic released a new games launcher which they proudly described as vibe-coded. They boasted of significantly faster launch times compared to Steam, adding, Is this bad boy faster? Yes. Will it have some bugs? Yes. Do we want to hear about them? Hell yes. And they sure have heard about them. Gamers posted a ton of visual glitches that made it seem like zero quality control went into the release, but at least the speedup was real. Xbox, meanwhile, is taking a different approach, spinning off— creating an internal lab to experiment with AI. Of all the AAA studios, Xbox has been particularly hard hit over recent years. Making this something of a fork-in-the-road moment. Many industry pundits thought the brand might be shuttered or sold off by Microsoft after years of poor performance. In a memo to staff this week, CEO Asha Sharma discussed the many emerging challenges of the moment. She specifically mentioned the wave of decompiled games being rebuilt to run on different hardware or in the browser. Sharma wrote, these experiments challenge long-held assumptions about our industry: the tether between software and hardware, the boundary between creation and play, and how people discover and experience games. While it's hard to predict the future of gaming, it's easy to see the possibilities in making great games easier to access, giving old games new life, and allowing more players to become creators and co-creators. These are unmistakably good outcomes for players, even when, and sometimes because, they challenge our existing business models. In response, Xbox has created Xbox Labs, A small internal skunkworks to, quote, turn emerging ideas into things our players and developers can experience. The plan is to build in the open and extend ideas made possible with emerging AI technology to deliver the future of gaming, whatever that ends up looking like. The response got a strong positive reaction, at least from the AI community. MIT's Christian Catalini wrote, a masterclass in what to do when AI tears up the rules your business was built on. Your games are being decompiled and run on hardware you don't sell. You can call the lawyers, or you can start a lab and embrace it. Summing up the bull position, Alex Finn wrote, gaming is the single biggest opportunity in AI right now. 2027 will be an absolute gold rush for anyone who dedicates the time to build a game with AI. AI allows solo devs to build the AAA games of their dreams with very little resources. He noted that while the gaming industry had up to this point completely rejected AI, as he put it, the first domino fell today with the introduction of Xbox Labs. Another interesting example of legacy companies adjusting to the AI era is Bumble's billion-dollar take-private deal. On Monday, the Wall Street Journal reported that Ithaca Holdings was in advanced talks to take ownership of the struggling company. Ithaca is a recently founded investment firm helmed by former Sequoia partner Ravi Gupta and former Meta Chief Revenue Officer John Hegeman. On Tuesday, Gupta announced the deal was done, writing, over the past 6 months, we've looked at hundreds of companies, asking which businesses would become more valuable as AI advances. Bumble stood out. As intelligence becomes more abundant, we believe connecting with the right person and forming meaningful relationships will become even more precious. And frontier technology can help more people find and build those relationships. Our purpose at Ithaca is to apply frontier technology to things that matter. We acquire enduring businesses, transform them with technology, and hold for the long term. We focus on businesses not only for their financial potential, but their ability to enrich lives. Now, Bumble has been struggling to make it work over recent years. The company peaked at a $14 billion valuation following their 2021 IPO, but the stock price collapsed by 95% in the following year. Gen Z's rejection of dating apps turned one of the strongest consumer software categories into a complete dud over just a few years. VC Sarah Guo is leading the investment and cheering on the takeover thesis. She wrote, consumer internet companies at scale are some of the most fun to be had, and a tiny crack team has never been better set up to run them. This team is fast, knows scale and business quality, and is extremely AI-pilled. With full ownership, they can rebuild whole companies around it. This is another concept that's been around quite a bit over the last couple of years— basically an AI-flavored twist on the private equity strategy of taking struggling businesses and giving them a tech makeover. Earlier this year, Italian holding company Bending Spoons went public with a mandate to refurbish defunct tech brands using AI. Their stable includes AOL, Vimeo, Meetup, and Evernote, and last month they added Airtable to the mix, excluding of course the new HyperAgent product that spun out from the company. The thesis is that these brands still have revenue and a viable product, but overheads are dragging them down and their futures involve inevitable change. By aggregating staff and using AI to reduce maintenance costs, Bending Spoons believes they can squeeze better performance from these legacy brands. Bumble obviously isn't the exact same situation, but Ithaca clearly believes they can resurrect and reinvent the brand for this new AI age. Speaking of Bending Spoons, Berkshire Hathaway has delivered a massive vote of confidence in the AI transformation strategy. Warren Buffett's investment firm has agreed to invest $330 million in private placement. The market was delighted with the move, sending the stock soaring by 40% on Wednesday. Berkshire was already a large investor in Bending Spoons, putting $600 million into the company during their July IPO, and this new round of investment takes their stake to 4%. The injection of cash gives Bending Spoons a new war chest to pursue another round of acquisitions. Then again, some companies remain focused on simply bolting on agents to allow them to expand their business and offer new services that wouldn't have previously been possible. One unexpected example is David's Bridal harnessing AI to offer full-service wedding planning. For 76 years, David's exclusively sold wedding dresses, and they sold enough to become the largest wedding dress retailer in the country. Beginning last year, they began building a new strategy to transform the business into an AI-powered wedding planning service. The strategy has been dubbed Aisle to Algorithm, and CEO Kelly Cook said it was aimed at growing David's in an asset-light way. Cook said that this was a huge bet that needs to pay off, Commenting, this is a 180-degree pivot. The company filed for bankruptcy in 2023, citing a weak recovery from the pandemic, and the brand was bought by asset management firm Scion Investment Corp. Now, obviously, the wedding dress business hasn't been disrupted by AI, but a secular downtrend in marriage over multiple decades has left David's looking to take a larger slice of the wedding industry. Overall, it seems to be a pretty extensive overhaul. David's has built an AI-powered wedding planner for brides that connects to a vendor portal, Essentially the standard marketplace play with an agentic twist. President Elena Vilck said that David's big advantage is the integration of the AI platform with customer data. She said, we have the actual brides coming into our system. When they book their appointment, when they commerce with us on davidsbridal.com, when they go through our surfaces, they're sharing a lot of information that is unique to them. In the tech stack, David's is using a combination of different models to power their agent and plan to add support for Meta's Muse soon. AI has also made a previously costly software play more manageable. Vilk said the platform was built with standard agentic coding tools and a fairly lean team, commenting, what would have taken 100+ people before, we were able to do with just tens. There's no way we could have achieved anything like this even 3 years ago. What's interesting about all of this is that the backdrop of it is yet another jump in the model capabilities. Opus 4.5 at the end of last year brought agents to the fore, and the combination of Astra and Opus 5.5 are once again, it seems, unlocking new opportunities. As always, it is worth being very cautious about prognostications of rapid change, and yet there is clearly a sense among many of these legacy players that change is now the name of the game and they have to evolve to survive. There will inevitably be pain along the way, but a lot of interesting new opportunities as well. So with all these new superpowers in your agent, what games are you going to create? Maybe we'll have to do that for an AI Operators Bonus at some point, but for now, that's going to do it for today's AI Daily Brief. Appreciate you listening or watching as always, and until next time, peace. Peace.