America’s trillion-dollar bet on AI

Recently, President Trump hosted an unusual dinner for two at the White House, dining alone with Dario Amodei, CEO of artificial intelligence company Anthropic. Why the sudden tete-a-tete?

Because Amodei sees existential safety issues with AI and wants to slow its development. He has called for government regulation. Trump has called the safety concerns “a hoax.” He opposes any slowdown or regulation.

Why is AI suddenly so consequential and urgent? Almost overnight, AI has become the beating heart of our economy. Realistically, AI-spurred high GDP growth is the only way we can meet our national debt obligations without drastic spending cuts or onerous tax increases.

While the two highest-profile AI companies, Anthropic and OpenAI, are the essential core of this new industry, they are not fully integrated enterprises. Most AI economic activity is being generated by a large constellation of other players — mainly hyperscalers, which have already invested more than $1 trillion . Given this very widespread activity, the centrality of AI to the economy can best be discerned from government economic statistics.

According to the Bureau of Economic Analysis , in the first quarter of this year, investment in the AI-dominated industry category of computers and peripherals contributed about one-quarter of the 2.5 percent growth in real GDP.

After slowing in the second quarter, likely, it rebounded in the third.

According to the Atlanta Federal Reserve Bank’s GDPNow estimate , real GDP growth accelerated to a 3.6 percent annualized rate in the third quarter. While GDPNow does not break down domestic private investment, overall nonresidential fixed investment accounted for nearly 1.6 percent — almost half — of that growth.

With some good reason, investors see AI as the modern equivalent of the Industrial Revolution.

Anthropic, which was founded only five years ago, has been expecting to raise $100 billion in an initial public offering, valuing the company at an eye-popping $2 trillion, according to The New York Times. That’s an extraordinary valuation, since Reuters reports that Anthropic had only $8 billion in net operating income in 2025 .

Now AI faces a triad of sudden threats: safety concerns, resistance to construction of huge AI data centers, and a risk that it is somehow a bubble that can burst like the dot-com bubble at the turn of the century, dragging down the whole economy.

While AI faces these new threats, our ginormous national debt continues to grow.

We hit $40 trillion in U.S. Treasury debt (gross debt) in August, according to the Monthly Treasury Statement . Net of Treasuries in government accounts that is at about $32 trillion — pretty close to current GDP . National debt last hit that level of GDP at the end of World War Two.

To dig out of our debt hole, GDP growth has to exceed growth in the debt.

Since the economy and net debt are now the same amount, the analysis is straightforward. In broad terms, the debt grows by the amount of the deficit, which can only be closed by new borrowing.

The deficit in fiscal 2025 was $1.8 trillion and beginning debt was $28.3 trillion at the end of 2024 , yielding 6.3 percent growth in the debt. Estimated nominal (comparable) GDP growth was only 4.8 percent, according to the Congressional Budget Office . This past fiscal year, debt grew 6.7 percent , driven by a deficit of $2.0 trillion and beginning debt of $30.3 trillion. The Congressional Budget Office estimate of nominal GDP shows 4.9 percent growth. Debt has been growing faster than GDP.

Yet, in the just-ended third quarter, real GDP growth was the aforementioned 3.6 percent, which when combined with general inflation of 3 percent or more produces a fast pace of nominal GDP growth of 6.6 percent or more. This would suggest that GDP can outgrow the debt.

Next year, however, will be much more challenging, since September brought a surge in interest rates, which will increase interest costs and widen the deficit significantly.

Already, net interest has been very growing rapidly, more than doubling from $425 billion four years ago to $1.1 trillion in the fiscal year just ended, as low-rate Treasuries issued after the Great Recession and during the COVID era have been maturing and replaced by new higher-rate Treasuries. This past year, one out of every seven dollars of federal spending went to pay interest, according to the Monthly Treasury Statement .

The current fourth quarter will be especially challenging, since Uncle Sam receives the least amount of tax revenue — and, therefore, has to borrow the most — in this period each year.

As a nation, we have a daunting balancing act to perform: we must exercise sufficient control over AI to ensure that it doesn’t go rogue and create chaos or worse, while still nurturing AI’s explosive economic potential to help avoid an otherwise almost certain fiscal-financial crisis. Like it or not, we are all-in on an enormous AI bet.

Red Jahncke is president of the Townsend Group.