US ‘has to win’ the AI innovation race, OpenAI exec says
His comments came after President Joe Biden signed an executive order to lease federal land sites for the construction of new data centers.
“So what you get with the Biden administration today is — at least from a signaling perspective — on federal land, trying to short the timeline between when you can get your project shovels in the ground and then the project going forward,” he said.
Lehane said the incoming Trump administration sees AI through two lenses — national security and economic security — and that he hopes both sides of the coin will coalesce into a national strategy.
He also said the input ingredients that make up AI infrastructure –– chips, workforce, energy and data –– are critical to the U.S. maintaining its leadership in AI, particularly against China.
“So my hope, our aspiration, is that you're going to get an overarching strategy that is going to come at it to deliver on those two goals, those two outputs, and with the urgency that this moment demands, because this is a zero-sum game; the U.S. absolutely has to win this,” Lehane said. “Winning is going to dictate whether the world is built on free, democratic AI rails, or whether it's going to be built on authoritarian, autocratic AI rails, and we have to win that.”
What Lehane hopes to see in future executive action and policy is work with states — both on and off public land — to expedite scaling out and building up AI infrastructure. He said that incentives for states to actively participate in this will help spur the process, such as sharing compute power specifically for state uses.
“I think some of the challenges around data centers have been: Would the states really benefit from this? They get some construction jobs; they get some operations jobs; they get some revenue. Those are all good things. But can the states actually begin to think about these data centers as being larger AI ecosystem economic development opportunities?” he noted.
Lehane said this approach would help motivate states to stand up AI infrastructure quickly, which could then be used for state-level projects.
He also noted that an interest in codifying this type of infrastructure programming into one or more pieces of legislation could further solidify incentives for state governments to help scale out domestic AI infrastructure.
]]>