Michael, thank you for being here today. I'm very excited to do this interview. You have had quite the career. So, you are now director of OSTP. For those who don't know, that's the Office of Science and Technology Policy in the Trump administration. You cover a wide range of technology: AI, quantum, nuclear energy, semiconductors, drones. So, just a little bit of responsibility, but we're going to focus on AI. Selfishly, because I'm an AI reporter and because that is just the hot topic right now. You've actually been thinking about AI policy since the pre-ChatGPT days when you were in the first President Trump administration, also working at the OSTP as CTO. You also have a background in the private sector working for Peter Thiel. You were most recently leading policy at AI startup Scale AI before returning to government. So I'm curious: what have you learned from your Silicon Valley days, and why are you excited to go back into government now?
Yeah. So I started my career in Silicon Valley in 2010, and I was with Peter and that team for seven years until joining the first Trump administration. And there was a sort of interesting, changing face and nature of venture investing over those seven years. Those were times when we were investing in SpaceX, Spotify, Airbnb, Stripe. And as more and more of our companies came across our desks, what we would always be seeing was that the amount of time and energy that we were spending in thinking about the regulatory risk associated with those particular investments or how government or policy more broadly would affect those became a bigger and bigger part of our investment thesis. And to me, I'd always been passionate about policy. And to me, I deeply believe that policy actually plays a very, very important role in how quickly the United States can continue to lead and accelerate and win in these core technologies. Whether it's drones doing commercial delivery, whether it's supersonics flying again, whether it's nuclear power plants being approved by the NRC, these are all sort of regulatory decisions that have to be made by government. And you can have the greatest technology in the world that you've invented in your garage or even attempted to scale, but in many cases, unless the government gives you approval, you can't actually commercialize those. And as a country, we definitely have to keep winning in these technologies. And the government has an important role to play in making sure we can unlock this for the American people.
Any lessons that you learned from Silicon Valley that you're trying to bring to government?
Well, to me, I think in Silicon Valley, when I was in those years looking at a wide variety of technologies, we were always trying to look over the horizon and figure out what's next. What's the next big thing that we have to be thinking about and getting ahead of? And that's a lesson you definitely have to take into government. I mean, government is certainly not known for moving quickly or being particularly agile, but especially when it comes to creating a regulatory environment which supports and encourages innovation in America, the people working on it have to be looking around the corner. They have to understand that a new technology is happening. And the way that a current regulation is written or the way that a statute is currently drafted, there's going to be a problem that's coming around the bend. And it's really important that you get in front of those and make sure it's not a problem. I'll give you one sort of minor example we think about a lot is supersonic flight. With supersonic flight, the current rule on the book is a speed limit. It is illegal to fly over Mach 1 over the continental United States. Now, there have been companies out there that have demonstrated what they believe that they can fly over Mach 1 without creating a sonic boom. Now, if that's actually true and that it doesn't create a noise issue by flying over Mach 1, that rule is totally antiquated. So at some point, someone will be able to create a commercial supersonic airplane that doesn't create a sonic boom, and the regs just aren't there to allow it. So that's just a minor example, but it extends along a wide variety of technologies.
So there are many prominent tech community leaders who have gone on to advise in this White House administration. David Sacks and of course Elon Musk probably being the most prominent right now. Just today we're seeing a fallout sort of happen between Musk and the president. What do you make of that?
Well, I can't speak to that particular instance, but I will say generally, what the president is able to do is assemble an unbelievable coalition of Americans that are committed to ushering in a golden age of American innovation. We're at a point in American history where we have the choice as Americans and as a government to bring these new technologies to bear for the benefit of the American people. And what's so unique about this particular election versus ones many years ago is that folks who may not necessarily have been considered to be on the same team a few years ago are all now united behind a president to drive this innovation for the country. And that's why I go to work every day, and that's why great patriots like David Sacks work alongside me to get that done. And we're excited for the team we built.
All right, let's move on to some other news of the day. There is a controversial provision in the current congressional funding bill to ban states from regulating AI for 10 years. We're seeing some bipartisan opposition to that particular provision. Recently, Representative Marjorie Taylor Greene and Senator Ed Markey have called for its removal, along with a big group of state lawmakers. What do you think? Do you think that the federal governments or the state governments should regulate AI?
So generally, this has been a problem with tech regulation for many years. I think probably the most famous fight in the preemption debate, if you will, was around privacy regulations. And for many years, there was lots and lots of attempts at the federal level to create comprehensive privacy legislation for the entire country, and there was unable ever to get to 60 votes in the Senate. And what ultimately happened is California kind of ushered in privacy legislation that very much mirrored what was happening in Europe, and then other states tried to do it. And what you generally tend to see is that there are significant downsides to a patchwork of individual state regulations. It makes it very challenging often for the smaller upstart players to be able to successfully commercialize across the United States because they have to comply with 50 different sets of rules. So I think generally, I'm a supporter in instances where federal preemption does make sense, so you can create an environment where all players, big and small, can succeed. I don't think the very large tech players have any problem hiring enough lawyers to make it all work at all the states. But I think for smaller players, preemption can make a really, really big difference in spurring innovation. So we'll see how that's playing out. From what I understand, the Senate is now working through an alternative version to what the House proposed in the bill that was passed last week. So we'll see what they come up with, and hopefully there can be a compromise between the two chambers.
So I want to talk about the change and sort of the tone of policy on AI from this current administration compared to the past. In recent months, it feels like we've moved from conversations about AI safety and the role of government to protect that more and more toward AI opportunity, in what I believe you've called the golden age of American innovation. So why do you think we're seeing this shift away from safety? And what do you say to those who remain concerned about the downsides of AI, everything from bias in AI systems to potentially catastrophic risks to humanity?
Yeah, it's been for those of you who track this intersection of AI technological progress and then the policy piece of it, the arc kind of began in the first Trump administration. President Trump was the first president in history to ever sign an executive order on artificial intelligence. That was in 2019, well before ChatGPT came out, where the president said this is one of the most important technologies for the future of our country and for the American worker, and we have to put a strategy in place to drive that forward. And he prioritized research and development, did a bunch of work on workforce development, and was able to promulgate the first-ever regulatory guidance on AI in 2020. What we saw over time was this optimism about how powerful this technology could be to help the American people, empower American workers, drive economic growth was, as we moved into the post-ChatGPT era, a lot of leaders who were in that LLM space ran to the Hill. And many of you may remember the tone of many of the early Senate and House hearings were very much about the risks associated with AI. There was this general infatuation with trying to identify what's the worst possible thing that could go wrong. And they went as far as even claiming maybe we should have a whole new agency or regulatory commission to oversee artificial intelligence. But over time, I think what the bipartisan consensus started to come into place that, in reality, artificial intelligence is a ubiquitous technology that's going to be touching all sorts of different applications. Whether you're flying drones, whether you're driving autonomous cars, whether you're developing AI-powered medical diagnostics, this technology is going to be part of that new technological breakthrough that you're making. So in a sense, from a regulatory front, the best way to actually approach this is to maintain kind of an industry- or sector-specific, use-case-specific approach to regulation. So the same regulators that are regulating vehicle safety should be regulating autonomous vehicles. The same people at FDA who are approving medical diagnostics should be approving AI-powered medical diagnostics, and you just sort of increase the bubble of that particular regulatory agency.
And I think on top of it, you had an administration in the prior administration that was sort of infatuated with discussing the fears or the problems that may come about of AI. And I think what this led to was a really negative tone around what this technology could provide, and ultimately set kind of a groundwork that, if it had continued, I think would have put the US in a real disadvantage to our global competitors. What we're trying to do in this administration is we understand that there will be changes that artificial intelligence will bring to the country. We understand that there's certain things that we need to monitor and track, but at the end of the day, we're in a global race and we need to be able to lead in that race. And as the president says, maintain unquestioned and unchallenged global technological dominance, and with that, that is how we can impart all the benefits of the American people and ensure that globally—because I know we may get to this—is we want the world to be running on the American AI stack. Vice President Vance was in Paris in the first couple weeks of the administration. He talked about the importance of this gold standard AI. We have the best AI in the world here in the US, and we need to make sure that we continue to lead on it.
So your mandate is to make sure that the US stays ahead on AI. With the launch of DeepSeek this winter, that was a wake-up call for everyone about just how close Chinese AI is to that here in the US. What are you doing to maintain that lead, and how are you doing that as you're also dealing with restrictions on student visas, on immigration that do impact the talent pipeline here in Silicon Valley?
