New Signs of a Bubble, and Trump’s AI Safety Zigzag

Covering AI security, infrastructure, and geopolitical risk.

It’s been kind of a weird week in AI, and in terms of markets, it’s feeling a bit toppy.

For one thing, Nvidia had a great first quarter, with revenues coming in at $81.6 billion, up 85.5 percent year-over-year. Demand is continuing to boom. And yet the company’s stock slid in the days after the announcement. So it seems like investors are not satisfied with merely excellent growth.

Meanwhile, SpaceX, which now owns xAI, has officially filed for its IPO, OpenAI is preparing for its own IPO, and OpenAI and Anthropic are said to be racing ahead with their own filings. Why the sudden rush?

There are perfectly rational reasons for these already well-funded companies to be pursuing access to deeper public markets. They’re facing hundreds of billions of dollars in future infrastructure bills, and they need to get ready. But why is it that three of the very top AI players just happen to all be pushing for IPOs at the same time? It may be that they see a peak to investor enthusiasm rapidly approaching, and they want to cash out appropriately. Maybe they’re looking down from the peak now.

Here it’s worth noting that Anthropic and OpenAI both internally have as-yet-unreleased superintelligent AI models that reportedly could disrupt financial markets.

That’s the financial side of things. Yes, maybe there’s a bubble, and maybe it’s about to pop. On the tech side, though, I don’t see any signs of a bubble. Progress appears steady and ongoing, in kind of a crazy way. And that may be part of the reason for another weird thing that happened this week: Trump’s abrupt cancellation of a planned AI safety event.

The ceremony was meant to mark an Executive Order creating a voluntary AI safety framework in which AI companies would preview their frontier models to government officials ahead of their public launches. According to at least one report, this would be a 90-day window, with part of the rationale being to help make sure the Pentagon is fully prepped for any cybersecurity impacts. But it seems there were internal disputes about the wisdom of this EO, and these have been communicated in only the sketchiest terms by President Trump.

“I didn’t like certain aspects of it. I postponed it,” he said to reporters. “I think it gets in the way of — you know, we’re leading China, we’re leading everybody, and I didn’t want to do anything to get in the way of that lead.”

As for what was really going on behind the scenes, the situation is a bit murky. Axios sourcing suggests that Trump and David Sacks, the Chair of the President's Council of Advisors on Science and Technology, just really don’t like regulation, and that Meta CEO Mark Zuckerberg and xAI CEO Elon Musk both urged him not to sign the EO. But Musk himself has denied this, stating on his X social media platform, “I still don’t know what was in that EO and the president only spoke to me after declining to sign.”

Amid all this, there has been some chatter among AI folks on X and Discord that Anthropic and OpenAI already have their next frontier AI models ready, and that they’ve been delaying their release in anticipation of some regulatory efforts from the White House.

So, it’s a weird situation. Maybe not weird for the Trump team: there is a bit of a history there with respect to abrupt pivots and messy comms. And maybe not weird for AI in general, since there’s been a pattern emerging in which AI capabilities bump up against state control in ways that neither side really knows how to navigate. But weird for the rest of us, in absolute terms, as we continue to watch this world-historical transformation unfold.

Alex Perala

Editor, Control Plane