$60 Billion and a Robot Walk Into a Bar: Big Tech's Wildest Week Yet

Lauren: You Derek: Welcome back to Tech Insider Weekly. I'm Lauren here with Derek. And okay, so get this. It has been a wild week in AI. Speaker 3: Wild is underselling it. We've got billionaires writing checks that have nine zeros in them like it's nothing. Derek: Casual Tuesday energy. So look, Forbes is calling it the largest startup acquisition ever. SpaceX just bought AI coding tool Cursor for $60 billion, all stock. Speaker 3: Wow. 60 billion for a coding assistant. Derek: Yeah, yeah, we're going to look at whether that price tag is strategic genius or organized panic. Speaker 3: I have thoughts. Strong thoughts. Derek: And then, plot twist, Jeff Bezos isn't sitting this one out. His AI startup Prometheus just raised $12 billion at a $41 billion valuation pre-product. Speaker 3: The phrase they're using is artificial general engineer, like not a chatbot, an AI that designs jet engines. Derek: That's a nice story. We'll see if the numbers back it up. it up? Speaker 3: There it is. Derek: We've also got Apple. Siri finally got its big iOS 27 overhaul. And Craig Federighi said, and I'm quoting here, Siri is not your AI girlfriend. Speaker 3: A sentence that needed to be said, apparently. Derek: And on the robotics front, a French startup just unveiled a wheeled robot, not humanoid. Backed by Eric Schmidt, the whole non-humanoid wave is getting interesting. Speaker 3: Interesting. Wheels over legs. There's a real argument there. Derek: All right, SpaceX, $60 billion, the biggest startup deal ever. Derek, what are we actually looking at here? So SpaceX just dropped $60 billion on a code editor. Speaker 3: Code editor for Rockets. Derek: For Rockets. Forbes reported it this morning. SpaceX is acquiring Cursor, the AI coding startup, in an all-stock deal. $60 billion. That's the largest startup acquisition in history. Speaker 3: Wait, back up. Like not just the largest tech, the largest startup acquisition ever. Derek: Ever. Full stop. Sandy Carter at Forbes confirmed it. Cursor wasn't even a household name outside of developers. Sold for Circles three months ago. Speaker 3: I mean, I use Cursor. It's good. But $60 billion good? Derek: That's the question, right? And here's what makes the number even harder to sit with. SpaceX exercised an option to buy. This wasn't a cold approach. They had a relationship, they watched the product, and they still wrote that check. Speaker 3: So this wasn't panic, or at least it wasn't impulsive panic. Derek: Organized panic, the most expensive organized panic in startup history. Speaker 3: Okay, but walk me through the logic, because on the surface SpaceX builds rockets and satellites. Why does Elon Musk need to own the tool his engineers type code into? Derek: So the CNBC piece from Ashley Capoot framed it this way. This is about competing with Anthropic and OpenAI, both of which already have popular coding tools. Claude writes code. GitHub Copilot, which runs on OpenAI, is everywhere. SpaceX just went public and Cursor slots directly into xAI and Grok. Speaker 3: Right. So it's less we need a better IDE and more we need to own a piece of the AI division. I develop our stack before our rivals lock it up. Derek: Exactly; and Cursor has real traction. Engineers at major companies, not just startups, are using it daily. You're not buying Sixty Billion dollars of vaporware. Speaker 3: Although Sixty Billion is a lot to pay for traction, like what's the revenue on this thing? Derek: That's the number nobody's publishing right now. The Los Angeles Times piece from Nilesh Christopher flagged the competitive pressure Speaker 4: Cursor's under. Derek: of pressure angle hard, Anthropic and OpenAI both have coding tools that are eating developer mindshare, but the actual Cursor revenue figures not out there yet. Speaker 3: So we're valuing a product with unknown revenue at $60 Billion because it has good vibes and Musk wants to beat Anthropic. Derek: Good vibes might undersell it, but yeah, the strategic logic is if AI coding becomes the default way software is written. It's written, and it's trending hard that way. Then whoever owns the best tool owns a choke point. Speaker 3: Wait, and this comes right after SpaceX's Wall Street debut? Fast Company noted that timing specifically-so you've got fresh capital, and suddenly you're swinging for the fences. Derek: That's a nice story-but show me the numbers. Because sixty billion in all stock means Cursor's founders and investors are betting SpaceX stock keeps climbing. that's not a cash out that's a bet on Musk's whole empire holding together right Speaker 3: Which has been a pretty volatile bet lately, depending on which week you're looking at. Derek: and in Engadget Igor Bonifacic pointed out this is specifically meant to bolster xAI's capabilities so Cursor doesn't just sit inside SpaceX's engineering org it feeds Grok it feeds xAI's broader ambitions Speaker 3: So you're buying a developer tool, but what you're actually buying... is the training data, the usage patterns, the feedback loop from millions of engineers writing code every day. Derek: Now you're getting to it. That's the real asset. Every autocomplete, every bug fix, every refactor, that's a signal about how humans solve engineering problems and that data at scale is what trains the next generation of models. Speaker 3: I mean, that reframes the price completely. $60 billion for a coding tool sounds absurd. Heard sixty billion for a continuously updated data set of how the world's engineers think? Derek: Still absurd, but at least you can draw the line on the whiteboard. Speaker 3: At least there's a whiteboard. Okay, so Musk is spending sixty billion to own the AI layer underneath software development; meanwhile, across town—or, I guess, across the billionaire stratosphere—someone else is making an even stranger bet, not about who writes the code, but who builds the actual physical thing the code controls? Derek: And that someone just raised $12 billion with a concept that makes AI coding assistance sound almost boring by comparison. Speaker 3: What does it even mean to build an artificial general engineer? Derek: Switching gears here, Bezos just raised $12 billion for something called Prometheus. Speaker 3: $12 billion for a company launched in November. Derek: November! And TechCrunch's Marina Temkin reported the valuation landed at forty-one billion before a single product ships. Lauren: SpaceX dropping $60 billion on Cursor, Derek: Bezos raising $12 billion in seven months. It's a big week for the number billion. It's a big week for the number billion. Lauren: It really is. Okay, so $41 billion valuation. Let's get real for a second. What are investors actually betting on here? Derek: So the phrase Bezos himself used, and this is the one that caught my attention, is Artificial General Engineer, not AGI. AGE]) Lauren: That distinction matters. Derek: It really does. The whole AGI conversation has been about software eating software, reasoning, language, code, but Bezos is pointing at atoms—physical products. According to The Verge's Emma Roth, the pitch is AI-powered tools to help develop physical things—jets, engines, medical devices, complex industrial systems. Lauren: So not a chat bot that writes your emails, a system that designs Speaker 4: your house. Lauren: designs a turbine. Derek: Exactly. And that's a fundamentally different problem. Software eats itself. You can train on code, test in simulation, iterate in milliseconds. Lauren: Mm-hmm. Derek: Physical engineering has the real world pushing back the whole time. Lauren: Which is why the valuation gives me pause. $41 billion for a company that launched in November, co-run by former Google executive Vikram Bajaj? Annie Palmer reported that. And the product is essentially a vision. Derek: Yeah, that's a fair hit; the New York Post noted Prometheus is targeting things like jet engines and medical devices, but the timeline on those is years, not quarters. Lauren: That's a nice story; show me the numbers. Derek: Right. To be fair, though, Bezos said publicly-CNBC had the quote-"We're not being secretive." He's actively trying to signal this isn't stealth mode for the sake For the sake of it. Lauren: Which is an interesting move. Usually you stay quiet until you have something to show. Derek: Unless the fund raise is the show. You announce the scale, you signal seriousness, you attract the engineering talent you need before a competitor does. Lauren: I mean, I get the logic, but twelve billion dollars buys a lot of credibility you haven't earned yet. Derek: Or it buys the time to earn it. Physical AI is slow. Drug design, industrial manufacturing. These aren't six-month sprint cycles. Speaker 3: Right Derek: If they're right about the category, the runway matters enormously. Lauren: And this is where Bezos' post Amazon identity gets interesting. He's not chasing the software A.I. race directly-OpenAI, Anthropic, that whole arena. He's betting the next unlock is physical systems, which honestly connects to what we're seeing in robotics this week too. Derek: Right, because if you're building AI for jet engines, you're not that far from building AI for factory floors. The Prometheus thesis and the robotics boom are pointing at the same Speaker 4: thing. Derek: pointing at the same thing. Lauren: The physical world is the next frontier. And right now, everyone is planting a flag. Derek: Forty-one billion dollars worth of flag. Lauren: At minimum. All right, from a $41 billion concept to something a billion people already have in their pocket, Apple dropped a major Siri overhaul this week, and the headline from Bloomberg was not exactly a rave. Derek: Just good enough to ease its AI crisis. I mean, I mean as endorsements go. Lauren: That one's going on the box. We're going to get into what actually changed, why it took so long, and what Craig Federighi said that surprised a lot of people. So Apple dropped its big Siri overhaul with iOS 18, and Bloomberg's verdict was just good enough to ease its AI crisis. That's a headline. Derek: That is a headline. Not exactly greatest comeback story ever told, right? Lauren: Right. And look, I've been waiting for this Siri update longer than I've waited for most things in my life. Derek: Same. So what actually changed? Lauren: Full app control, on-screen awareness, it can actually read your emails and act on them. 9to5Mac got a pretty detailed breakdown from Apple's Mike Rockwell on why it Why it took this long, basically they rebuilt the understanding layer from scratch. Derek: Which is actually the right call-you don't patch a broken foundation, you pour a new one. Lauren: Architecturally, yes, but users don't care about foundations, Derek. They care that Siri still mishears call mom to call Tom in 2025. Derek: To be fair, Tom deserves a call sometimes. Lauren: Tom has been waiting patiently. But, okay, here's what I actually can't get past. Craig Federighi gave this interview at WWDC and MacRumors covered it. He said, and I'm not making this up, Siri won't be your AI girlfriend. Derek: I mean, points for honesty: Apple is basically saying we're building a useful assistant, not a parasocial companion. Lauren: And I get the positioning. It's the dig at Character AI and the whole emotional attachment thing, but it also tells you something about where Apple thinks the market is versus where competitors are actually going. Derek: Right. Google and OpenAI are leaning hard into the relationship layer. Memory, personality, continuity. Apple is opting out. Lauren: Deliberately, which depending on your worldview is either principled product design or a massive gap in the market they're leaving on the table. Derek: Hmm, I go back and forth. Privacy first users will love it, but the people who actually use AI assistance daily, they want something that remembers them. Lauren: That's the bet Apple is making, that trust and privacy win over depth. Whether users agree, we'll see. Derek: Just good enough is not a rallying cry. Lauren: No, it is not. Okay, but now flip that on its head. All week we've been talking about AI that lives in software. Your Siri in your ear, cursor in your IDE, the next story is about AI that physically shows up to work, robots rolling into factories and the people betting that humanoids are not the answer. Derek: Oh, this one is wild. Let's get into it. Lauren: Switching gears, here's a take I didn't expect to be defending today: humanoids are overrated. Derek: Oh, I love this. Business Insider ran a piece on Genesis AI making that exact argument. Lauren: French startup, backed by ex Google CEO Eric Schmidt, and they unveiled a robot called Eno. Wheeled, not bipedal, the whole design philosophy is stop trying to make robots look like people. Derek: And you know what? The engineering case is actually pretty strong. Human oil Lauren: The point form is expensive to stabilize. Derek: Mm-hmm. Lauren: Two legs are a liability in an industrial setting. Derek: France 24 covered the reveal, commercial deployments are planned for late 2026, and they're partnering with LG for distribution. That's a real enterprise channel, not a demo reel. Lauren: Wait, LG? That's a wild pairing. Derek: Bloomberg reported it today. Schmidt-backed startup meets consumer elect... Electronics giant Industrial AI meets Global Supply Chain. Lauren: For the non-engineers in the audience, what LG brings here isn't brand recognition, it's manufacturing scale and existing B2B relationships. Eno gets distribution that would take a startup a decade to build. Speaker 3: Wow. Derek: And then over in Barcelona, The Cursor just closed a 73 million euro Series A, also non-humanoid. but also targeting industrial production-EU startups covered it Thursday. Lauren: 73 million euros at Series A for robots that don't even walk. Derek: Exactly. And Google DeepMind is piling in, too. They just launched a European robotics accelerator, 15 startups in the first cohort per Digital Watch Observatory. Lauren: Hmm, so DeepMind isn't just doing the research anymore, they want skin in the game on actual hardware deployment. Derek: That's the signal. When DeepMind runs an accelerator, they're not... They're not being charitable. They're creating an ecosystem they want to sit at the center of. Lauren: And that connects to something bigger this week. You had SpaceX buying Cursor for the developer stack, Bezos building Prometheus for physical engineering AI, and now the robot layer. Derek: It's the same race, different terrain. Everyone is trying to own a piece of the infrastructure that everything else runs on top of. Lauren: Code, rockets, warehouse floors, like you said. Derek: And the humanoid versus wheeled debate actually captures the real question. Do you build AI that mirrors human biology or do you optimize for the environment it's actually working in? Lauren: Turns out factory floors don't have stairs. Derek: Yeah, exactly. The make-it-look-human instinct might be the thing slowing everyone down. Lauren: What I want to know is adoption timeline, because we've seen a lot of robot demos that stay demos. Derek: There. Genesis says late 2026 for commercial deployment. The Keir has fresh capital to burn, and with DeepMind accelerating 15 companies simultaneously? The pressure to actually ship is real. Lauren: The demos are getting harder to hide behind. Money's in, timelines are public, enterprise partners are named. Derek: The week's theme basically wrote itself. Who owns the AI layer underneath everything—software, physical systems, the assistant in your pocket? Lauren: And we are very much not done asking that question. Derek: Okay, so rockets, code editors, atoms versus software, and Siri's very complicated love life. Not a bad week. Lauren: Honestly, the thread that kept pulling for me was the SpaceX-Cursor deal. Sixty billion dollars, all stock, largest startup acquisition ever, and the real asset isn't the tool. Derek: It's the data on how engineers think. Lauren: Yeah, exactly. And Bezos pointing at atoms with Prometheus, it's the same instinct. Whoever owns the infrastructure layer, software or physical, Derek: wins the longer game. Lauren: valuations are wild. Derek: The ambition might be real. real. Both things can be true. That's the week in a sentence. Lauren: All right. If you got something out of this, Derek: subscribe wherever you listen, subscribe wherever you listen, leave us a review, and tag us if there's a founder we need to sit down with. Lauren: New episodes drop every Wednesday. Thanks for writing this one with us. Derek: See you then.