Welcome back to the Technology newsletter 🤖
OpenAI is in full spring cleaning mode: shutting down products, ditching consumer experiments, and reshaping itself into an enterprise company ahead of what's looking like the biggest tech IPO in years.
Meanwhile, a leaked AI model rattled cybersecurity stocks, and the self-driving car revolution just hit a new milestone.
Let's get into it 👇
✍ A leaked draft blog post revealed that Anthropic has been testing a new AI model called "Mythos", described internally as a "step change" in capability that scores dramatically higher on coding, reasoning, and cybersecurity tasks. The leak spooked investors so badly that cybersecurity stocks like CrowdStrike dropped 7% and Palo Alto Networks fell 6%, on fears the model could find and exploit software vulnerabilities faster than defenders can patch them. (Fortune)
🤖 Google launched new "switching tools" for Gemini that let you import your full chat history and personal preferences from ChatGPT or Claude, basically making it as easy as possible to break up with your current AI assistant. You export a ZIP file from your old chatbot and upload it to Gemini. It's the AI equivalent of phone number portability. (TechCrunch)
💳 France's Mistral just secured $830 million in debt financing to build a Nvidia-powered AI data center near Paris with 13,800 GPUs. It's Europe's most serious attempt yet to build homegrown AI infrastructure instead of relying on American cloud providers. Mistral wants 200 megawatts of compute capacity across Europe by end of 2027. (CNBC)
📱 Apple is planning to open Siri up to rival AI assistants, including Google Gemini and Anthropic's Claude, in iOS 27 this fall. Right now, only ChatGPT is plugged in. The update will let you choose which AI service handles your Siri queries through a new settings menu, basically turning the iPhone into an AI platform rather than a locked-down walled garden. (Bloomberg)
🛞 Waymo just hit 500,000 paid rides per week across 10 U.S. cities, a tenfold increase from two years ago. The company has quietly expanded beyond its original markets of Phoenix, San Francisco, and LA into Austin, Atlanta, Miami, Dallas, Houston, San Antonio, and Orlando. Self-driving cars went from science experiment to actual transportation option faster than most people realized. (TechCrunch)
🔴 Disney pulled out of its planned $1 billion investment in OpenAI after learning (less than an hour before the public) that Sora was being shut down. The three-year deal would have let Sora generate videos using Disney, Marvel, Pixar, and Star Wars characters. No money ever actually changed hands, but it's a dramatic collapse of what was supposed to be AI's biggest entertainment partnership. (Variety)
OpenAI is doing something unusual for a company that spent the last three years launching products at a breakneck pace: it's shutting things down.
Last week, the company killed Sora (its AI video generator that launched just months ago) dropped its ChatGPT shopping features, and told employees to stop working on "side quests."
If you're wondering what's going on, the answer is three letters: IPO.
Sora's shutdown tells the clearest story. The app was burning through roughly $1 million a day in computing costs while its user base collapsed from a peak of about a million to fewer than 500,000.
When you're trying to convince Wall Street you can be profitable, a money-burning video toy isn't the look you want.
The collateral damage was real, too: Disney found out its planned $1 billion investment in OpenAI was dead less than an hour before the public announcement, ending what was supposed to be AI's splashiest entertainment deal.
The ChatGPT shopping experiment got a similar haircut.
Users just weren't buying things through a chatbot, so OpenAI is pivoting to "product discovery" instead of trying to compete with Amazon.
Behind the scenes, the reorganization goes deeper.
Fidji Simo, OpenAI's CEO of Applications, held an all-hands meeting in mid-March and told staff the company is "orienting aggressively" toward enterprise productivity — coding tools, business automation, the kinds of products that generate predictable, high-margin revenue.
Meanwhile, CFO Sarah Friar has been stacking the finance team with IPO veterans, including the former chief accounting officer at Block and the former CFO of DocuSign.
OpenAI also retained the law firm that handled Google's 2004 IPO.
None of this is subtle.
OpenAI is targeting a Q4 2026 listing at a valuation of around $850 billion, which would make it one of the largest public offerings ever.
SoftBank just took out a $40 billion bridge loan with a 12-month repayment window to fund its $30 billion stake in the company (a bet that only makes sense if the IPO happens on schedule and goes well).
What we're watching is a company that spent years as a freewheeling AI lab transforming itself, almost overnight, into something that looks like an enterprise software company.
Whether that version of OpenAI is worth $850 billion is a question the public markets will answer soon enough.
What did you think of today's newsletter?