EXPERT INTERVIEW — World leaders and tech executives are gathered in Paris for the latest global summit on artificial intelligence. The French AI summit, co-hosted by India, is raising the need to ensure inclusivity and sustainability with the adoption of the technology. But those sentiments are coming under increased pressure amid the rapidly accelerating global race to innovate and lead the AI space – seen most recently in the sudden breakthrough of Chinese AI startup DeepSeek, and the Trump administration’s moves to repeal guardrails on the technology and boost investment in U.S. AI infrastructure. For his part, French President Emmanuel Macron announced a 109-billion-Euro AI investment ahead of the Paris summit, and said that Europe will cut back on AI regulation to “resynchronize with the rest of the world.”
Before the summit, The Cipher Brief spoke with Gilman Louie, a co-founder and early CEO of In-Q-Tel, the pioneering technology investment firm funded by the CIA, about the burgeoning global AI race. Louie suggested that the Chinese need to solve for a “scarcity” of compute abilities caused by U.S. export restrictions may have actually spurred the innovation at DeepSeek. He added that such innovation will further push the “accessibility” of AI, and serve as one more catalyst in the AI race. Louie also spoke about President Donald Trump’s newly-announced U.S. sovereign wealth fund, which he said could be a key tool to fund AI development.
Louie spoke with Cipher Brief CEO Suzanne Kelly. Their conversation has been lightly edited for length and clarity. You can also watch the discussion on our YouTube channel.
Kelly: We've had some time to digest the initial reaction to DeepSeek. What is your take now that we've had a chance to learn a little bit more about this in terms of how much of a threat this is to U.S. national security?
Louie: We have to start at a high level of how we got here in the first place. The U.S. advantage has always been its technical base and the use of certain tools that we had in national security to restrict flows of technologies overseas, particularly to a place like China. And so in the prior administration, the focus was on restriction of capabilities, particularly around microelectronics. On one hand, the U.S. tech companies, and four or five major AI companies, are in the San Francisco Bay Area. If you look at what took place, it was, OK, we have the chips, they don't. We have the latest generations, the B100s, they have the H100. They have limited supplies and we don't, and we can build our foundational compute capabilities around abundance. They announced a half trillion dollars going into new server farms and new capabilities — and that was just one cohort of companies. So that was seen as our ability to keep the advantage.
Now what's interesting is, as anybody in engineering knows, scarcity creates a different kind of innovation. So to some of us, it wasn't a surprise that another group of talented engineers and scientists could come up with a different cut on the problem — to solve a class of compute with far less compute. It’s not that different than in national security, when the U.S. had huge computational advantage, so the Soviets had to invest in algorithms; they became very good at creating algorithms because they didn't have access to compute. Fast forward 60, 70 years later, and the same thing just happened in China.
If we had a different environment where they didn't have the scarcity, they might not have gotten to that solution. And because we have abundance, there is no incentive here to solve for scarcity.
Kelly: So the scarcity of the materials they needed to launch this is really driving more innovation in China?
Louie: Yes, particularly on things like algorithms and mathematical approaches to go off and solve these classes of problems. Now the reality is, the current innovation of DeepSeek, at least from an algorithmic point of view, is a momentary advantage. Any great engineer, they can look across the way and say, you may not show me your trick, but I know it is now a solvable problem. We have smart people who will solve that problem. And you combine that with open source, it means that the performance gap will level itself off.
The markets panic because it's, My god, we don't need as many NVIDIA chips – dump NVIDIA, we don't need all these server farms. Let's go in the opposite direction. But whenever there's been a computational breakthrough in performance or price, name a moment in history when that resulted in less compute. The demand for compute will go up because of this. What this will result in is accessibility. In the old world, if you weren't one of the big five AI companies, it was very difficult to see yourself competing on world-class applications if you didn't have billions of dollars of investment to compete against them. DeepSeek has shown that you don't need that. In some ways, you're getting a democratization of AI innovation because of this new price-performance breakthrough. And it's coming from China, it's not coming from the U.S., to the point where the former CEO of Intel, Pat Gelsinger, announced in his new AI startup that he's going to use DeepSeek. The former CEO of Intel, which enjoyed billions of dollars in investments from the U.S. CHIPS Act on the argument of national security, now is going to benefit from using technology built in another country, because of scarcity.
The intersection of technology, defense, space and intelligence is critical to future U.S. national security. Join The Cipher Brief on June 5th and 6th in Austin, Texas for the NatSecEDGE conference. Find out how to get an invitation to this invite-only event at natsecedge.com
Kelly: The last time we spoke, we were talking broadly about this tech race between the U.S. and China, particularly when it comes to national security. I asked you, what are going to be the indicators as to whether we were ahead, or we were behind? The first thing you said was, watch the commercial market. How do you feel about that today?
Louie: The speed with which AI companies and application companies are growing — like 30 a day — that's going to start accelerating. Now it's a slightly different race, and this is a real dilemma for the United States: the bias is going to be a bias for action. Prior to this recent set of breakthroughs, everybody was worried about AI security, AI safety, ethical issues around AI. Are we going too fast? Should we slow down? You had petitions of people saying we need to go slower, we need to be more thoughtful. Countries, nation states, whole alliances like the European Union who say we need to have conferences. The French want to lead this discussion. The U.S., even the Department of Defense, we want to have the discussion.
Now because of accessibility, there is definitely a race. That race means that if you go too slow, using the old models and the old techniques because you're inhibiting innovation, the other team may burn right by you. And the next time around, you may not be able to catch them. So you're going to see a logarithmic increase of AI-powered applications around large language models using this new framework that DeepSeek has, whether you use DeepSeek or somebody has an alternative version of that. That's going to accelerate growth, because it's available and it can be everywhere. And that will change how we think about where AI computes. If you need smaller models, less computation, that means that potentially you can do more computation on the edge, which also changes the power equation quite a bit, if I don't have to go to the cloud.
Kelly: President Trump ordered the creation of a U.S. sovereign wealth fund. You were a CEO and co-founder of InQtel – all those years ago, the CIA realized it could get into the venture capital market as a way to spur innovative ideas for solutions that would help the government with a lot of the problem sets they had. I really wanted to get your thoughts on how these sovereign wealth funds work in general, and what they are created to intend to address.
Louie: With In-Q-Tel, it's hard to imagine it took place 25 years ago. It made us think differently about how venture capital can play a role in national security.
This new announcement is an important one. It's the realization that globally, sovereign wealth funds, the ability to leverage anywhere from a half a trillion to $2 trillion to invest in innovation, is a new engine that fuels innovation. And if you look at the U.S., we've been very advantaged in the last software revolution. The next technical revolution around frontier technologies is going to require deep tech investing. Deep tech investing requires a different kind of investment profile, and you have to have much more patient, limited partners. Limited partners are people who invest in venture funds. Sovereign wealth funds are a method for nation states to support that set of activities. The U.S. is at a huge disadvantage when you compare it against the Netherlands or Saudi Arabia or the Chinese or even a country like Singapore, which has more than one fund.. Venture funds listen to that International LP base more than they listen to the U.S., simply because the U.S. is not in the game.
If the U.S. is going to win and maintain its innovation advantages, it needs to find a method to provide the necessary risk capital to invest in things that allows America to build again — not just innovate, but to actually build. That entire supply chain of infrastructure requires a different kind of investment.
Now there's two parts of this. The difference between a sovereign wealth fund and an initiative is that initiatives are putting money into the public sector or the private sector, and we hope that would stimulate growth. We've put a lot of money in through stimulus in the last handful of years. The reality is you're giving money away. There's nothing that comes back. What other nation states have realized is if you use the sovereign wealth fund model and you do it right — you actually have professional managers who know how to manage funds and you invest in good funds and good projects and programs — they'll have a return on investment. And that return not only pays back the original investment, but it sets the table for future generations. So it is a tool that the U.S. needs to have. And I think the Trump administration rightly identified that this is a different way of allocating resources to fund innovation that ultimately could have a payoff, not just in keeping the U.S. ahead in the technology race, but actual economic benefits for future generations of Americans if these funds are invested properly.
Everyone needs a good nightcap. Ours happens to come in the form of a M-F newsletter that provides the best way to unwind while staying up to speed on national security. (And this Nightcap promises no hangover or weight gain.) Sign up today.
Kelly: What else is top of mind for you as you look forward? There's been a lot of very fast change.
Louie: A few things. Let's start with the Department [of Defense].
Given the speed of innovation, particularly the capacity of a country like China, we have to rethink not only the tools that we use if we get ourselves into a fight, but how we build them. The U.S. has always had, since World War II, a manufacturing advantage over our adversaries. You take a country like Germany in World War II, they were more advanced than we were, but they did not have the capacity to manufacture at the rate necessary to sustain a long fight. The U.S. did.
If we get ourselves into a bad situation with a nation state like China, and hopefully we do not, our ability to sustain is more important than just the shiny weapon that we're going to bring into the fight. We're learning that in Ukraine right now. In order for us to do that transformation, it means that there's going to be pain in the department because we have to stop doing some programs. We can't continue to spend at the rate that we're doing. We're going to be spending closer to a trillion dollars in this next budget on the Department of Defense. And the question is, with all that money we're spending, while we have certain strategic and tactical advantages over our competitors, our competitors are quickly catching us and in some areas surpassing us, in things like hypersonics, and we need to rethink that equation.
And then I think from an intelligence point of view, AI changes everything. Yes, AI is going to change autonomy and defense quite a bit, and how we think about warfare. But how we understand what other nations' intents are – either against the U.S. or to understand how they think about the U.S. – the old model of very sequential tasking and collection and processing and dissemination, that model was built for a different time in a different era, when we were organized around tools of collection rather than tools of knowledge. And so the IC [intelligence community], if it wants to maintain its relevance in the world, not only needs to do a better job in sensing what's happening around the world, but it needs to be able to process it in a way where the information gives us information advantage, decision advantage, whether it's a war fighter, the president of the United States, a principal, or whoever we're supporting in the IC. We need to rethink the equation.
Read more expert-driven national security insights, perspective and analysis in The Cipher Brief
