OpenAI unveiled its latest policy blueprint at an event hosted by the Center for Strategic and International Studies in Washington, D.C. on Wednesday.
The blueprint details how the United States should maintain its leadership in artificial intelligence and meet the enormous energy demands required by AI technology.
In the document, OpenAI describes an optimistic vision for artificial intelligence technology, calling it a “foundational technology like electricity” that can provide numerous jobs, GDP growth, and investment opportunities, while offering “an unmissable opportunity for reindustrialization.”
Of course, it includes a hook that’s hard for almost any American to resist: “Reviving the American Dream.”
▲Source: PYMNTS
At Wednesday’s event, OpenAI’s Vice President of Global Affairs, Chris Lehane, said they spent “significant time” discussing AI infrastructure needs with both the Biden administration and Trump’s team.
Trump plans to repeal Biden’s executive orders on artificial intelligence, claiming they “hinder AI innovation.” Additionally, Trump has acknowledged that the U.S. needs to expand energy supply to maintain competitiveness in AI and suggested loosening permits for fossil fuels and nuclear power use.
OpenAI has explicitly stated its willingness to work with a potential Trump-led new administration on AI policy.
▲Chris Lehane (Source: Q BERLIN)
Specifically, OpenAI’s policy blueprint proposes establishing a “North American AI Alliance” and developing a “North American AI Compact” to streamline access to AI talent, funding, and supply chains, while competing with similar initiatives proposed by China in artificial intelligence.
OpenAI indicates that this AI cooperation mechanism would start with the U.S. and its neighbors, then expand to America’s global alliance network, including Middle Eastern countries like the UAE.
▲Source: Economic Observer Network
To incentivize states to accelerate AI infrastructure licensing and approval, OpenAI also suggests that states and federal government jointly build “AI Economic Zones.”
Lehane points out that as the U.S. entered the digital age, much economic benefit flowed to coastal regions, while the relatively “lagging” Midwest and Southwest could become potential core regions for AI investment—these areas have land and capacity to build wind farms and solar arrays, with potential for nuclear facilities.
Given that the U.S. Navy operates about 100 small modular reactors (SMRs) powering submarines, OpenAI proposes leveraging Navy expertise to build more civilian small and medium-sized reactors to increase nuclear capacity.
▲Source: IAEA
Lehane also considers establishing a data center in Kansas and Iowa, states with “extensive agricultural data,” to create an agriculture-based large language model or reasoning model. These facilities could not only serve communities but make them “centers of agricultural AI.”
Reports indicate that “China built as much nuclear power capacity in 10 years as the U.S. did in 40 years” and continues to approve new reactors. Citing estimates, Lehane stated that by 2030, the U.S. will need 50 gigawatts of energy to support AI industry demands and compete with China.
And facing this competition, “we have no choice,” Lehane said.
▲Source: Pixabay
Furthermore, as existing programs cannot keep pace with AI-driven demands, OpenAI anticipates introducing a “National Transmission Highway Act” to expand electricity, fiber connectivity, and natural gas pipeline construction, while seeking new powers and funding to remove obstacles in transmission planning, permitting, and payment.
Notably, the EU also released yesterday its first draft “code of conduct” for General Purpose AI (GPAI) models, outlining risk management guidelines and providing a blueprint for companies to comply with regulations and avoid severe penalties.
As countries worldwide continue to advance AI infrastructure implementation and promote AI technology standardization, how should China “respond”? Particularly, how can we maintain our advantages in policy, data resources, and application scenarios while addressing shortcomings in basic theoretical research and computing power industry?
Solving these issues should be our trump card when choosing to “accept the challenge” in the face of competition where there is “no choice.”