Key Takeaways
The US is squandering its lead in artificial intelligence, with a fragmented national response that risks ceding the most consequential technology of the century to China, according to a sweeping proposal from Axios co-founder and Chief Executive Officer Jim VandeHei.
Starting next year, American AI companies will spend roughly $1 trillion annually — a sum equal to the combined inflation-adjusted cost of the Manhattan Project, the Apollo moon landings, the Interstate Highway System and the Human Genome Project. Yet the country lacks the coordinating infrastructure, workforce planning and safety protocols that the moment demands, VandeHei wrote in a Wall Street Journal op-ed published Thursday.
"America faces one of those rare, historic moments when government, business, schools and families could be working together to meet a truly generational challenge: winning the AI race," VandeHei said. "So far, we're blowing it."
The spending surge is concentrated among a handful of frontier labs — OpenAI, Anthropic and Google DeepMind — whose engineers are pushing the boundaries of what AI can do. Anthropic recently built a model so capable it decided not to release it publicly over safety concerns, a decision that arrived with essentially no government preparation or public debate, VandeHei noted. Meanwhile, China has designated AI a national strategic priority, aligning its top universities, state companies and military around a single development road map with spending at a scale akin to wartime mobilization.
The stakes extend far beyond corporate profits. AI will determine who controls space and warfare, dominates the global economy and shapes the first technology with super-intelligence surpassing human capability. The US approach — treating AI primarily as a business story about rich companies with sci-fi ambitions — is dangerously incomplete, VandeHei argued.
A National Project, Not a Technology Story
VandeHei's seven-point plan reframes AI as a shared national project akin to the World War II mobilization, the Marshall Plan or the Apollo program. The first step: establish a coordinating body drawn from the federal government, leading AI companies, business and labor, economics, public health and ethics — people paid to anticipate rather than react haphazardly. This group would map potential problems and upsides before they hit, build response playbooks in advance and communicate honestly with the American public.
The second priority is preparing for job displacement before it becomes a crisis. VandeHei proposes a staged response plan with automatic triggers — if unemployment hits 6 percent, prebuilt solutions including job training programs or temporary mechanisms to slow layoffs would activate. He also suggests new forms of communal, human-centered work funded by the enormous wealth AI could generate: a nurse corps, an eldercare corps, a tutor corps and a community rebuilding corps that address the loneliness epidemic, the collapse of local institutions and the teacher shortage.
A national AI labor app would pull real-time data from AI companies on where they need data centers, energy infrastructure, chips, engineers and technicians, then match workers with jobs and appropriate training. Meta and Google have already funded small versions of the training piece, VandeHei noted.
Safety, Education and a Global Coalition
The proposal takes future dangers seriously. Recursive self-improvement — where AI teaches itself and works autonomously — could be months away, VandeHei warned, a threshold that could produce superhuman capabilities and potential rogue behavior. Biosecurity risk follows the same logic: as models become more capable, the barrier to engineering dangerous pathogens drops. In each case, the risk is known now, and preparation remains optional.
On the global front, VandeHei calls for a US-led coalition built around American technology, models and rules, with a worldwide AI supply chain and safety protocol. Member countries would gain access to the most powerful economic and technological alliance in the world, more dynamic than the United Nations or NATO, while developing nations would be enticed to align with the US rather than accept Chinese infrastructure deals.
"The window to do this hasn't closed. It has narrowed," VandeHei wrote. "China can impose that kind of coordination by decree. They can align government, industry and society toward a single technological objective through instruments that are unavailable and undesirable in a democracy. That's not an option in the US. We would need to make a different choice."
For investors, the implications cut both ways. The $1 trillion annual spend benefits AI infrastructure providers including Nvidia Corp., Microsoft Corp., Amazon.com Inc. and Alphabet Inc., whose data center and chip businesses stand to capture a significant share. Nvidia trades at roughly 35 times forward earnings, reflecting market expectations that AI capital expenditure will continue accelerating. But the regulatory and safety risks flagged in the proposal — potential export controls, safety-driven deployment delays and workforce disruption — could introduce volatility. The absence of a coordinated national strategy, VandeHei argues, is itself a risk that markets have not priced in.
This article is for informational purposes only and does not constitute investment advice.
