When the government funds high-tech research, should it be for true moonshots? That’s the new question surfacing about the Biden administration’s biggest, most lavishly financed tech policy. The debate is centered on a new entity called the National Semiconductor Technology Center, an R&D project hatched in the 2022 CHIPS and Science Act — which Congress made the centerpiece of its $11 billion longer-term strategy to make the U.S. more competitive in the future trillion-dollar microchip industry. Much of the CHIPS money, so far, has been for manufacturing plants and job creation. The more future-facing part — research and development to unlock the next generations of microchips — is just ramping up now. The Commerce Department kicks off a search this month for potential locations for the NSTC headquarters and two technical centers, which a long list of states including New York, California, Arizona, Oregon, Texas have been jockeying to land. So if America is about to spend billions of dollars to accelerate R&D, what should those efforts really do? The options range widely, from exploring entirely new materials to providing startups with access to lab equipment to experimenting with different approaches to AI hardware. A handful of experts have begun arguing lately that the real payoff will be looking very far forward, and using the center to pursue high-risk projects — the kind that might fail, but might also remake how the whole industry operates. A key proponent is Jordan Schneider, a podcast host and think tank expert on the U.S.-China tech rivalry, who co-wrote a widely discussed post making the case for the paradigm-shifting moonshot approach. He and other microchip policy thinkers, including“Chip War” author Chris Miller, have been arguing that is the best way for the center to pay off and distinguish itself globally. They draw the comparison to the Defense Advanced Research Projects Agency, which helped transform chipmaking by providing seed funding a generation ago to far-reaching technologies that overlooked at the time by corporate road maps. Those early investments in design software, computing architectures and chipmaking tools drive the chips currently powering artificial intelligence. Today, the semiconductor industry faces a whole new set of societal grand challenges — questions around sustainability, energy use, and workforce shortages — that are relevant for not only chipmakers but the future of computing-intensive technologies like AI. Central to Schneider’s argument is the funding strategy behind the NSTC. The law will provide it with money for five years. Then, unless Congress intervenes, it will have to transition to a self-sustaining organization paid for by semiconductor companies themselves. Schneider argues that the clock is already ticking: the NSTC should pursue bold, concentrated bets now, while it is still being subsidized by the government. Its largest potential corporate funders may be less excited about a risky research agenda. “Companies are not going to be super-excited about the type of research that has the potential to blow up their business models,” Schneider said. That’s just the type of research that can pay off for the American economy as a whole, however. The moonshot-based strategy would also resolve a dilemma for the NSTC: its need to differentiate itself from already established international centers for semiconductor research, most notably the public-private powerhouse IMEC in Belgium. “IMEC is cool, but if we’re spending $11 billion to make a knockoff IMEC, we’re wasting our time here,” said Schneider. Which way is the Biden administration leaning? It’s a little soon to tell. The early project announcements seem to be more about identifying interesting, relatively low-cost projects for the center to ease into. Coming this summer are competitions for $100 million in hypertechnical topics, such as AI-based design of radio frequency chips and test vehicles. Schneider said a more conservative approach might be more appealing at the start: “They’re going to want to have wins which are happening on a shorter time horizon than a 15-year vision to create chips out of DNA.” But there are recent signs that the effort has much bigger ambitions. It recently hired Maryam Cope, the top Washington lobbyist for Dutch chipmaking equipment company ASML, to lead its government outreach. Cope told POLITICO that she will press officials at all levels to support the NSTC's work toward R&D that is “not being done anywhere else in the world, such that you’ve stimulated a semiconductor R&D renaissance in the United States.”
| 
Comments
Post a Comment