As the tech industry prepares for a sea change in Washington, its new power players in artificial intelligence are expanding their policy footprint to match. OpenAI, at the white-hot center of the AI boom, is no exception. Chris Lehane, the political veteran and former Clinton White House lawyer who now leads OpenAI’s global policy, sees an expansive collaboration between industry and government in coming years as not just Washington, but state and local governments adapt to the dawn of powerful new AI tools. The company is circulating an as-yet-unpublished “blueprint” for AI infrastructure, with proposals for special AI economic zones and cross-North American policy collaboration. Political observers will note that its ideas are carefully tuned to the moment. The AI industry is about as “coastal elite” as they come — with high-paid workers clustered tightly in San Francisco and a couple of other coastal cities. But OpenAI’s new proposals have a distinct “rise of the rest” feeling, trying to frame goals around a geographically broader idea of growth. “Who benefited in the last 30-plus years from the digital economy? It disproportionately flowed to the coasts, and large chunks of the country have not necessarily fully benefited from it,” Lehane told DFD during an interview Tuesday. The “rest of the country” is often connected to the tech industry more as a power source than a talent pool, and Lehane suggested that energy might be a starting point for local AI-centric tech economies. “Those sections of the country … have a really unique opportunity to think about, okay, let's make sure we're not just building energy centers, let's think about how these energy centers can help us grow and support broader [tech] ecosystems.”
DFD spoke with Lehane about how industry and government alike can avoid the shortcomings of past attempts to build “tech hubs” in left-behind parts of the country, how the company sees itself working with a tech-forward Trump White House and how energy policy and AI policy could work in tandem. An edited and condensed version of the conversation follows: When you think about a state launching something like an “AI strategy,” where do you start? How do governments make sure they aren’t just throwing money at a feel-good project that never takes off? For these things to really work, there have to be a couple of elements in place. First of all, each time you do this, it does have to be different. It does have to match up to the particular technology, and then you have to figure out what your local competitive advantage is.
For example, if you're going to build a data center or some type of ecosystem in Detroit, it should be organized and built around how it would become the leading player in thinking about transportation and automobiles. There's so much data, information and expertise there, you're leveraging one of your existing advantages. That was one of the things that Pittsburgh did, it already had some great academic institutions, Carnegie Mellon, Duquesne and the University of Pittsburgh, so they were able to leverage a competitive advantage that naturally existed.
Locals need to lead on this. The locals are going to have as good of a sense of what will work, as opposed to it being jammed from the top down. One of the things that we’ve suggested is that you should take a chunk of the compute that's going to get created, and then let local governments sort out how they want to use it. If you’re looking at Detroit, Michigan does have some incredible, globally leading educational institutions. Could you take some of that compute and work with the University of Michigan, Michigan State [University], to stand up what could be an incredibly powerful ecosystem?
Should states think about their own energy strategy for AI? I gave a little talk last week at CSIS about infrastructure, and I drew an analogy with “Guns, Germs and Steel” by Jared Diamond. The thesis of the book is that countries that do really good jobs of marshaling their resources to get a competitive advantage are the countries that end up being successful.
For this “intelligence age,” what we’re thinking of as the AI period, that is going to be some version of chips, data and energy. Everyone who's going to want to benefit from AI is going to really have to think about energy. I'm an optimist, and I try to be a realist here as well. Some estimates are suggesting we're going to need 50 gigawatts of energy by 2030. But because there is such an imperative to do this, to make sure that U.S.-led, democratic AI prevails over a Chinese-led authoritarian AI, that's a geopolitical market force that's going to drive the need for energy. That creates the opportunity to
accelerate an energy transition, and even above and beyond that a catalyst for potential re-industrialization. Are you optimistic that under the next presidential administration there will be enough space for states, under a federalist framework, to experiment with whichever energy source they choose in scaling up to power AI? I am. Energy, broadly defined, does generate bipartisan support. West Virginia’s [Independent] Sen. Joe Manchin and Wyoming’s [Republican] Sen. John Barrasso had an energy bill that I could see some version of reemerging, even in the lame duck, but I think more likely in the next Congress.
Second, there are just plenty of sources of energy. The challenge in this country is not necessarily the energy sources themselves, it’s how we deal with the permitting stuff around it to make it happen. Third, at some level as a consequence of the last election two weeks ago, one of the big messages, particularly from my party, the Democrats, is that we need to transition from process into building. For both parties a big takeaway is that the American people want to see this country get back to doing what it does best, which is building. Energy is a natural place to do this.
You take all of those things, and then you take into account the geopolitics. If you’re looking at chips, data and energy, China has built-in advantages on the data. They can just scoop up all the data they want. They have a built-in advantage on energy. They just in the last year had 10 or 11 nukes that came online, and another 10 or 11 coming on next year. We lead on the chips, we lead on the innovation. But to maintain and continue that lead, we're going to need to do the energy piece.
Ultimately we're going to have to start to think about data too, in a creative way that’s consistent with protecting people's rights. But the energy piece ends up being the one that you can see all the parties and interests sort of coming around together. Are you planning on staffing up on policy for the coming years, and this next administration? We view AI as transcending partisan politics. It's not Republican AI, it's not Democratic AI, it's American AI. We began talking about this even as far back as July, that it's absolutely critical that U.S.-led AI prevail over PRC-led AI particularly when you think about values like democracy, freedom and opportunity. That's sort of our organizing principle in terms of how we think about how we engage.
| 
