This is an interesting and unusual one. There’s so much talk in the tech media right now about global outcomes, about a potential slowdown, and conflicting ideas on the future. But what about the day to day bartering that goes on between companies – and nations? What about a national government getting involved in how the private sector of the nation sells AI to other world governments?

For a little of that, we can turn to an interview that took place at the Imagination in Action event The Next Endeavor in Mountain View, CA, in September. (Disclaimer: With Imagination in Action, I help to put on these events.) This was a ground-breaking event, and very relevant to what we are all, techies and otherwise, facing right now. It’s a moment of great interest (often expressed as fear) and a time for introspection. So it was good to get a little inside scoop from Brandon Remington, the Executive Director of the U.S. National AI Center, who was interviewed by Nico Enriquez, Principal at Future Ventures, about this.

First, Remington noted that this is a big job, since effectively, the landscape is the entire world.

“How do you begin to sort that into something tangible and useful?” Remington asked rhetorically.

One piece of the puzzle is the technology stack. In response to Enriquez’s questions, Remington noted that the stacks are defined, and there’s some order to the ordering process. Enriquez asked about potential conflict between a corporation’s interest and the interest of the government, and about safety in general.

“Where something is dangerous, then we put in certain restrictions,” Remington said. “Where it’s dangerous to everybody, we put in more. Where it’s dangerous in only certain actors’ hands, we put in (guardrails) there, and we just try to scope it. Now, AI in particular, as we all know, is moving so fast, and the technology is changing, the understanding is changing. So it’s very much a continuous, almost daily process to get that right.”

It’s a competitive game, and Enriquez asked Remington about how that games out, given the reality that China has the edge on robots, and the U.S. has the edge on frontier models, but the Chinese seem to be developing lower cost solutions.

“There’s certainly the price of the object, of the particular use case,” Remington replied. “There are the dependencies, the security implications, the other values that are there, and then, like, what does the 5-year, 10-year outlook look like? I think people recognize: never bet against America and innovation. There’s a reason why America leads in this technology. There’s a reason why America leads in, you know, all kinds of innovation, and betting on the American tech stack is a good move.”

Enriquez also asked Remington about his work on Maven, an early U.S. government AI tool. Remington gave us some of the history, from a personal perspective.

“I was a lawyer in New York,” he explained. “I was asked by the Marine Corps to come, on orders, to the Pentagon and build AI. And I said, ‘I’m a lawyer. Surely you have someone better than this.’ And they were like, ‘We don’t have any AI experts. Please go help.’ So then I went there and I, you know, got to retool and be hands-on. And as it turns out, when you get to build technology, it’s very hard to go back to a law firm.”

Of current operations, Remington mentioned solutions like a data and context officer, and AI evaluations, and other protocols to ensure alignment.

Would the U.S. government, Enriquez asked, get involved in safety conversations as necessary?

“Applying a rigorous technical foundation to that, applying rigorous testing to know where it works, where it doesn’t, where the risks are is always important,” Remington said.

In answer to questions about what the U.S. needs to win on AI, Remington said it’s sort of like a company in some ways, in terms of what will work best.

“We need secure data,” he said. “We need to have all the controls that anyone else would. The things that are challenging: we have usually more bureaucracy. I’ll go head-to-head with anyone on bureaucracy. We got a lot of that. But we’re actually quite decentralized. Like, I can write my own AI policy for my boundaries, and change that around, within some broader constructs. So there’s actually, I would say, you really see federalism within this. And there are some really fast-moving parts of government. There are some slower ones, and it comes down to the ingredients that I would say anyone here who’s looking at a large company would find familiar.”

I’m also going to use some of Remington’s substantial quoting verbatim as he talked passionately about the goals of a newly created National U.S. AI Center.

“We’ve been slightly in stealth mode, but the goal here is a one-stop shop,” he said. “If you want to find out what the U.S. government is doing on any given thing, come in. Let’s talk. I can connect you with anyone. If you want to sell abroad, we can do that. If you want to sell to the U.S. government … then we can do that. If you want to talk about standards and safety and security, and any of that kind of stuff, we can talk about that. We’ll have the Patent Office in there. If you want to talk IP, we can do that. The goal is: we know D.C. is far. We know it can be hard to navigate. And the goal is to at least have a place where people can come and find this out, and we can connect you in.”

Here are some of the closing thoughts on AI sovereignty for nations, what that looks like, and what many of the client nations are expressing right now.

“They are very willing to lean in and be very thoughtful about the policies that they need for their country, and also, what is best situated for their country,” Remington said.

And, he suggested, it’s going to grow.

“You’ll see announcements from big players around very expensive infrastructure buildouts, if that makes sense,” he continued. “You will see, also, announcements from smaller players. And I will say, you’ll see a ton of follow-on activity, because once you’ve got that ecosystem there, you’ve just unlocked another market with how many millions more people, and then you can sell in all the app layers and a whole lot of other things right into that.”

With all of this in hand, Enriquez had two more questions. One was: what keeps you up at night?

“Making sure we execute,” Remington replied. “You know, back to, like, the boots-in-the-dirt thing. My focus is: take these high-level statements, make them real. What are all the things that we need to do to make that real? And those are all the things you don’t get to see.”

Enriquez also asked Remington one thing he would change about the U.S. government. That got an even shorter answer.

“Less bureaucracy,” Remington said.

And they agreed to have a beer.

As I say, this covered a lot of interesting ground in geopolitics. We have companies like Nvidia leading, not to mention Google, OpenAI and Anthropic. The international angle is a real one. Stay tuned for more.