The Fable 5 saga continues..

Mike Belsito unpacks three AI stories that matter to product builders this week: SpaceX's $60bn acquisition of Cursor and what the end of model neutrality means for your team's tooling; why Noam Shazeer joining OpenAI is a signal about where the next frontier of AI capability might come from; and how the Anthropic Fable 5 export control situation escalated all the way to the G7.

We discuss
— Why SpaceX's acquisition of Cursor is a product story, not just a finance story — and what the collapse of model neutrality means for developers inside someone else's platform
—Who Noam Shazeer is, and why his move to OpenAI signals that fundamental capability gains may still lie ahead
— The Fable 5 export control timeline — from launch to G7 summit — and what it means to build on promises that depend on a third party keeping theirs

Referenced
Cursor: https://cursor.com
Anthropic: https://www.anthropic.com
Attention is all you need (2017): https://arxiv.org/abs/1706.03762
Character.ai: https://character.ai

Transcript

SPEAKER_00 0:00

I'm Mike Belcito, and this is Now Shipping, the weekly AI news show for product people. Every week I'm bringing you three AI news stories that matter to you as somebody who's actually building products for a living, not the hype, not the noise, just the stuff that you need to keep up with. Brought to you by the team at Mind the Product, this is Now Shipping. Let me give you a quick preview of this week's three stories. Story one, SpaceX just paid $60 billion for cursor, the AI coding tool your engineers probably used this morning. And it might sound like a finance story, but this is a product story too. Story number two, a man named Noam Shazir just joined OpenAI. Now, you may not know that name, but if you've ever used Chat GPT, Claude, Gemini, or any modern AI model at all, you've used his work. We'll talk about why that's important. And story three, we're back to Anthropic. Anthropic promised subscribers 13 days of free access to Fable 5. Most got three. The free window expired just a couple days ago. The model's still offline. This week the CEO of Anthropic ended up at the G7 Summit negotiating with President Trump. So those are the three stories. Let's get into it. All right, story number one SpaceX and Cursor. Now, look, when I saw the headline, SpaceX acquires Cursor for $60 billion. My first instinct was to sort of put this in the, you know, interesting finance story pile. I mean, it's a lot of money. It's a huge deal. Congrats to the founders of AnySphere, the company behind Cursor. But for product people, actually, there is a lot going on here for product people. So let me set the scene. So Cursor, if you're not familiar, it's one of the fastest growing tools in the developer world. By February of this year, it had reached $2 billion of ARR. That was after 18 months. This is the fastest any business software company has hit that mark. And just in the past four months, it more than doubled. So the thing was growing so fast, kind of hard to believe. And then SpaceX, they just acquire it for $60 billion. Now it was an all-stock deal, but of course, SpaceX just went public. So the stock has real value. It will be liquid. It's the largest private company acquisition of a venture back startup ever recorded. Now, what does this mean for product people? Well, the reason Cursor was so popular wasn't just that it was a great coding tool, it was also a neutral coding tool. You could use it with anthropics models, open AI's models, Google's, whatever you wanted. It worked across providers. That neutrality gave it a kind of trusted, sort of independent status. It was like the Switzerland of AI coding tools. But now that status is gone. Cursor is moving into XAI's stack. They're already building a shared AI model together, one that will run inside of Cursor and inside Grok. And there are early reports of something called Origin, a new code repository platform being developed by the combined entity. And that's going to compete directly with GitHub. Now, think about that. SpaceX going after the code repository business against Microsoft, because of course Microsoft owns GitHub. Now, what does this mean for product people? A few things. First, the tools your team relies on are no longer neutral, right? Cursors acquisition, it's just the most visible example of a pattern that's being accelerated all year. The tools in a developer's daily workflow are getting captured by the front teal model players. And when a tool gets captured, the model neutrality goes with it. The implicit bet that you were making when your team adopted a tool that it would remain independent, that it would serve you regardless of who's winning the AI model race, that bet sort of just changed, right? Like you may want to ask yourself, which tools in your stack are actually still neutral anymore and which ones are becoming someone else's platform play. I mean, also, this is platform risk kind of personified, right? It's not just a tool conversation. If your engineers are deep in cursor, custom workflows, integrated context, built-up muscle memory, the switching costs just got higher. And the company that just acquired it has a fresh AI model, a code repository platform in development, and access to some of the most powerful computing infrastructure in the world. They're not building a coding tool, they're building a developer platform. And once developers are inside a platform, they tend to stay. So you want to understand what you're inside of because it's kind of hard to leave. And then I think it's also interesting to just watch out for Origin. Now, look, we don't have a lot of details yet, but if SpaceX and Cursor are building a code repository to compete with GitHub, and again, that's owned by Microsoft, you now have two of the most powerful companies on the planet potentially fighting over where your team's code actually lives. I mean, the last time that platform battle played out, the implications were enormous for everyone building on top of either ecosystem. So watch this one pretty closely. We'll see what happens here. All right, that's story one. Let's move in to story two. Okay, story number two, Noam Shazir joins Open AI. Um, I guess I should ask you a question. Have you ever heard of Noam Shazir? Well, if you haven't, you've still been living inside of his work every single day because Gnome Shazir is one of the eight co-authors of a paper published in 2017 called Attention is All You Need. That paper introduced the transformer architecture, the thing that GPT, Claude, Gemini, essentially every major AI model in the world is built on. You can make a very serious argument that no paper in the history of computer science has had a faster or more direct impact on the technology that people actually use. And this week, Noam Shazir joined OpenAI. Now, Sam Altman called it a hire that he had wanted since the very beginning of Open AI. Now, before I get into what this means, I'll share a little context if you're not familiar with Noam Shazir and his career arc, right? So after co-authoring that paper at Google, um, he was at Google when he authored that, he spent years working on large language models at Google. Then in 2021, he left to co-found character.ai, an AI platform focused on personalized character-based AI interactions. Now, Character became a major consumer AI app, had millions of users, they had real revenue. But then earlier this year, Google effectively acquired the key people and licensing in a deal that pulled Shazir back in. So that was their way of getting him back on Team Google officially. And then now he leaves Google and chooses OpenAI. And here's what makes it interesting: this is a man who had the full resources of Google. I mean, I do think it's a signal about where open AI thinks the next frontier is. Now, Shazir's specific enterprise has been in model architecture, the deep foundational stuff that determines what AI can and can't do. Not necessarily product, not go-to-market, architecture. Um, so when you make a hire like this, it's because you believe the biggest remaining gains are at the architecture level, potentially, right? Um, and that means, if that is the case, that OpenAI thinks that there are fundamental capability improvements still ahead, not just incremental model polish. Now, for product teams building on top of these models, I think that matters a lot. The capabilities you'll have access to in 12 to 18 months might be way different than what you're working with today. Not just, you know, faster and cheaper, but actually different. Um, and that could affect how you design towards AI native experiences right now. Um, also, I think it's a reminder of how fast the talent landscape moves in AI. Shazir co-wrote the most important paper of the decade in the world of AI. He went to found his own company, watched it get bought up and folded back into Google, and then he makes the choice to leave relatively quickly after getting acquired to move on to open AI. Now, that whole arc happened in four years. The people that are shaping the foundational technology that we're working with in, they move fast. And the companies they choose, again, it's a big signal about you know which companies are making something real and who, you know, at least they think are going to win the race. Um, I also think it's gonna be super interesting to see now what open AI builds next. That is story number two. We have one more to go. All right, story number three. Yes, we're back to Anthropic and Fable V. I know we covered it last week and the week before, but things keep changing, things move fast. It's just the nature of this AI everything world that we're in. Um, and yeah, so like where are we now? Well, I guess let's rewind a little bit. June 9th, Anthropic launched Fable V, their most powerful model ever. Um, subscribers basically got a 13-day free access window. Then on June 12th, three days later, the U.S. government issued an export control directive. The model had to go offline globally. Um, everybody was speculating everything. There was it was a mess, right? Then, June 20th, Anthropic's CEO Dario Amodi was at the G7 Summit meeting with President Trump to discuss the export control situation. It goes to show you how important AI is in conversations all over the world. Like it's a big geopolitical issue now at this point. This conversation was all about whether Anthropic and their most powerful model could be turned back on. Now, as of today, Fable 5 is still offline. There has not been an announced return date. The other thing is this week, the 13-day free access window, that expired already, right? So that's come and gone. So with the model down, most subscribers got between like either zero, one, two, three days of actual access of the 13 days that everybody was promised. Um, what does this all mean? I guess right now, first, like AI models are geopolitical assets, right? This isn't just a tech product. The fact that a CEO of an AI company was at the G7 table this week with the president of the United States, it tells you something about how governments are starting to think about frontier AI. This is a strategic asset, and that means the rules governing them are going to come from places that product teams have never had to pay attention to before. I mean, now you got to think about international trade law, national security directives, export controls. I'll be honest, I didn't even know what an export control was when it came to tech, but this is the world that we're building in now. I also think that the pricing trust story is something that, you know, we may want to pay attention to. I mean, Anthropic made a promise to their customers that everybody would have 13 days of free access, and then they couldn't keep that promise. Now, they didn't really have control over that, but I don't know. You were promised a feature and the government took it away, the clock ran out. Maybe we'll get it back. I don't know. I mean, Anthropic is offering credits and refunds, they're they're handling it. But the question for you as a product builder, have you made promises to your users that depend on a third party keeping their promise? I mean, this week, this is a clear example what happens when that chain sort of breaks. I mean, the liability ultimately lands on Anthropic, but it could land on you. So I think it'll be interesting to see how this gets resolved. Whatever deal gets struck, it does seem like a deal's coming at this point, but whatever deal does get struck between Anthropic and the U.S. government, um, it's gonna set in an interesting precedent. Uh will the government do this again to another frontier model? Um, will frontier model providers start building geography-based access controls into architecture? I mean, will international product teams start facing sort of like a two-tier AI world where the U.S. gets one set of capabilities, but maybe in other parts of the world gets another set of capabilities. I don't know the answer to this yet, but it doesn't seem like hypotheticals. It seems like questions that are relevant right now. And the answers to those questions will ultimately impact how you build. Fable 5 will be back online. Maybe it's next week's episode. We talk about how it's back online and none of this matters, but um, this is where we are this week. So that's a wrap on this week's episode of Now Shipping. If you found this valuable, uh please tell a friend, spread the word, every little bit helps, subscribe, leave a comment too. Like, look, if you want to see different things in this show, if you want different kinds of stories, um, anything like that, leave a comment. I will read your comments. I promise you, I want to make this even more valuable for you. So, with all of that, once again, my name is Mike Belcito and brought to you by the team at Mind the Product, this is now shipping.