Today in Tech & AI: The Real Conversations Behind the Headlines

President Donald Trump abruptly canceled a planned executive order signing on May twenty-first. The order would have given the government authority to test frontier AI models before release, but the AI CEOs invited for the signing… simply declined to attend. In our last episode, we talked about an AI model solving a mathematical problem that had stumped humans for eighty years, a story about pure capability. This week, the conversation shifted entirely to who actually controls that capability. The other threads that dominated the timeline this week were all about power, just measured in different ways.

Anthropic’s valuation surged to nine hundred billion dollars after closing another thirty billion in funding, and in the same week, they hired legendary researcher Andrej Karpathy away from OpenAI. The conversation wasn’t just about the money, but about what it signals when top talent makes a move like that. Meanwhile, Airtable’s founder Howie Liu announced his new Hyperagent program is deploying ten million dollars, not in cash, but in inference credits. He’s subsidizing five hundred founders to build “agent-first” companies, and the chatter on builder Twitter is all about what a company fundamentally looks like when an AI agent is the first… employee.

And in the most unexpected collaboration, the Vatican announced it’s releasing an encyclical on AI ethics, developed with tech leaders including Anthropic’s own co-founder, Chris Olah. It seems the company with the nine hundred billion dollar valuation is also the one advising the Pope. Each of these stories lit up its own corner of the internet, but the two that reveal the most about where we are right now are the White House showdown and the Anthropic money machine. Let’s start with the canceled executive order. The surface-level story is simple: Trump planned a signing event, the CEOs of the major AI labs were invited, they said no, and the event was called off.

The chatter online immediately split into two camps. The first was the obvious political take: Big Tech is now so powerful it can make a sitting president blink. The optics are undeniable. But the more useful conversation was happening on the threads between AI builders and policy wonks. The proposed order would have mandated a ninety-day window for the government to test new frontier models for safety and cybersecurity risks before public release. The AI companies had been pushing back, arguing for a fourteen-day window. Their argument, which echoed across developer forums and X, is that a ninety-day delay is a death sentence for innovation in a field that moves at lightning speed.

More importantly, they argued the government doesn't even have the expertise or infrastructure to meaningfully test these models. The sentiment was less "we don't want oversight" and more "you don't know what you're doing, and you'll break our momentum." Then the story got more specific. Reports surfaced that Elon Musk and Mark Zuckerberg had personally lobbied the president to cancel the signing. Musk immediately went on his own platform, X, to post a denial, stating the claims were "false" and that he wasn't even aware of the order's contents. This is where the story gets clearer. You don't need a conspiracy.

You just need to listen to what Trump himself said afterwards. He was quoted in Ars Technica saying, and I’m paraphrasing here, that he felt the order could have been a blocker. He said, "We're leading China, we're leading everybody, and I don't want to do anything that's going to get in the way of that lead." That’s the real story. It’s not about a secret cabal of CEOs. It's about a publicly stated, bipartisan consensus that the AI race against China is paramount. The AI labs successfully framed safety regulations not as a matter of public good, but as a threat to national competitiveness.

As Lizzi C. Lee from the Asia Society Policy Institute pointed out in a widely shared thread, both the U.S. and China are grappling with this exact same tension: how do you govern this technology without ceding your lead? The White House didn't back down because of a scheduling conflict. It backed down because the industry successfully argued that in the race for AI dominance, the only thing worse than a dangerous model is a delayed one. Now for the money. If the DC story was about political power, the Anthropic story was about financial gravity. The company's valuation hitting nine hundred billion dollars is a number so large it feels like a typo.

For context, that’s nearly triple its valuation from just February. The immediate reaction on VC Twitter was, of course, disbelief. How can any company, even a leader in the hottest sector on earth, be worth that much? But the smarter conversation wasn't about the headline number. It was about the mechanics underneath it. The best thread on this came from SaaStr, where venture capitalist Rory O’Driscoll laid out the bull case. He argued that every time Anthropic has raised money, its valuation multiple has been significantly lower than the multiple for a typical, high-growth Series A or B startup.

But Anthropic’s growth rate is far higher. So his argument is that, relative to the risk and the growth, it’s actually been the best trade out there. You’re betting on a near-monopoly in a world-changing technology, not just another software company. Then you have the Salesforce detail, which grounds it. Salesforce plans to spend three hundred million dollars on Anthropic tokens this year. It sounds massive. But as Jason Lemkin pointed out, that’s only about four percent of Salesforce’s total engineering wage bill. It’s a huge check, but it’s a measured bet on making their thousands of engineers more productive.

Lemkin’s killer insight was this: "Salesforce is spending $300,000,000 a year on Anthropic. Anthropic is probably spending … $3,000,000 a year on Salesforce. Both customers of each other. But one has 100x higher the TAM." TAM, total addressable market. Anthropic’s product—intelligence itself—can be sold to every company on earth. That’s the logic that gets you to nine hundred billion. And finally, there's the Andrej Karpathy piece of the puzzle. Karpathy is one of the most respected AI researchers in the world. He was a founding member of OpenAI and led Tesla's AI team. His move to Anthropic is a massive signal.

In the world of frontier AI, talent is the most constrained resource. When someone like Karpathy decides your lab is the place to be, it validates your research direction and your potential in a way that money alone can't. The valuation is the financial bet. Karpathy’s move is the technical bet. This week, they both landed on Anthropic. So what does this all set up? For months, the conversation has been about what AI can do—solve problems, write code, create images. This week, the conversation shifted to who gets to decide what happens next. We saw the political power of the labs in their ability to reframe a national safety debate into a conversation about geopolitical competition.

We saw the financial power required to even compete, with numbers that are starting to look more like sovereign wealth funds than venture capital rounds. And we saw the power of talent to signal which project is most likely to win. Last week's breakthrough in mathematics was a demonstration of the technology's potential. This week was a demonstration of its power structure. The question is no longer just what the models are capable of. It's who is capable of controlling them.