No AI FDA coming
Former White House AI adviser Sriram Krishnan basically poured cold water on the idea of a federal AI licensing body. His message: the Trump administration isn’t interested in building a giant bureaucratic gatekeeper for models, and it doesn’t want to start picking winners and losers in the AI race.
That matters because a licensing regime would be the kind of thing that makes startups sweat and hyperscalers groan. Think fewer “move fast and break things” vibes, more “please wait while our lawyers review the model for the third time.” Krishnan said that kind of setup would put “sand in the gears” of innovation.
The backlash isn’t going away
Krishnan also argued that the growing opposition to AI data centers is coming from the industry’s own messaging problem, not just policy. He pointed to local pushback against roughly 75 U.S. data center projects worth about $130 billion in the first quarter of 2026, which tells you the political heat is not imaginary.
- More public skepticism around AI could still slow permits, construction, and power deals.
- But the absence of a formal licensing regime is a win for speed, scale, and capital deployment.
- In other words: fewer federal shackles, same messy public debate.
Why investors should care
The big takeaway is that Washington may stay loud on AI, but it’s not necessarily about to become a bottleneck. That’s good news for the companies racing to build models, chips, and data centers — and bad news for anyone betting on a quick regulatory moat.
Krishnan also backed export controls that briefly paused Anthropic’s Mythos model after a security issue flagged by Amazon, which is a reminder that government oversight may stay selective rather than sweeping. Big picture: AI isn’t getting a full-speed green light, but it also isn’t getting a brand-new federal cop on every corner.