AI growth budget deficit: why Trump’s tax cut promise will fail
Republicans claim AI will supercharge growth and fix the budget. History and mathematics say otherwise.

Key Takeaways
- Republicans promised AI growth would generate enough tax revenue to pay for tax cuts, repeating a Reagan-era claim that has never worked
- Even 4.4% annual growth over a decade would only cut the deficit to 3% of GDP; balancing the budget would need 7.2% annual growth, which AI has never delivered
- AI companies face their own crisis: they need 45% annual revenue growth for seven years just to break even on $1.43tn in datacenter spending, diverting capital away from government bond purchases
The Republican promise that artificial intelligence growth will fix the US budget deficit, allowing tax cuts to pay for themselves, mirrors a claim that has failed repeatedly since Ronald Reagan. According to The Guardian, Treasury Secretary Scott Bessent is relying on AI to deliver 3% annual economic growth, which would somehow fill the government’s coffers and manage the rising national debt. History suggests it will not work.
The ai growth budget deficit problem is not new. Every time Republicans cut taxes since the 1980s, deficits widened, not narrowed. Yet the promise returns now wrapped in artificial intelligence optimism, with President Trump claiming the US will grow faster than ever before and use that growth to “take care of the 40 trillion” in federal debt. Financial markets are plainly unconvinced: the 10-year Treasury bond yield surged to its highest level in nearly 25 years, signalling investors are demanding extra compensation for lending to the government.
The immediate pressure on bond yields comes from war in Iran, which pushed inflation higher and forced the Federal Reserve to raise short-term interest rates. But beneath that sits a harder truth: the US government’s finances are unsustainable, and no realistic amount of growth will fix it on its own.
| Treasury secretary's growth target | Scott Bessent promised 3% annual economic growth, achieved only twice this century outside pandemic |
|---|---|
| Trump fiscal plan cost | One Big Beautiful Bill Act estimated to add $4.7tn to federal debt through 2035 |
| Required growth to balance budget | 7.2% annual growth needed by 2036; stabilising debt needs 2.5% productivity growth, hit only once since 1959 |
| AI datacenter investment needed | Superscalers need $13.1tn to $18.7tn additional revenue over next 10 years just to cover AI buildout costs |
| Interest payments as share of GDP | Rose to 3.3% from 50-year average of 2.1%, consuming more budget room |
| 10-year Treasury yield level | Hit highest in almost 25 years, signalling investors doubt growth promises |
Why the deficit math does not work even with AI
The Committee for a Responsible Federal Budget calculated the growth rates needed to hit different targets. To cut the deficit to 3% of GDP by 2036, assuming Trump’s temporary tax cuts become permanent as they have historically, the economy would need to expand at 4.4% annually for ten years. To balance the budget entirely by then would require 7.2% annual growth.
For perspective, the US has achieved 4.4% growth only twice this century outside the bounce-back from the Covid pandemic. The 7.2% rate is even more remote. The last time the US sustained that kind of expansion was in the 1950s and 1960s. To stabilise the federal debt relative to GDP alone would require total factor productivity (the efficiency with which the economy converts inputs into output) to grow 2.5% per year. The US has hit that once since 1959, despite electrification, the interstate highway system, the telecommunications revolution and the information technology boom.
These numbers matter because they show the scale of the claim. Trump’s fiscal plan, his One Big Beautiful Bill Act, is estimated to add $4.7 trillion to the federal debt through 2035. The deficit already stands at 6% of GDP, twice what Bessent once promised to deliver, and the Congressional Budget Office projects it will climb to 7% by 2033. Interest payments on existing debt now consume 3.3% of GDP, up from a 50-year average of 2.1%. That leaves less room each year for anything else.

Even booming AI would barely help the government’s books
Suppose, against the odds, that artificial intelligence does deliver transformational productivity gains. The desk’s read is that even a boom substantial enough to justify the venture capital being poured into it would do little to fix the budget crisis, for a counterintuitive reason: AI wealth would flow mainly to capital owners, not workers.
If AI automates cognitive work as enthusiasts hope, productivity gains would concentrate in the hands of shareholders and technology companies. The effective tax rate on capital gains and corporate profits today sits at roughly half the tax rate on wages and salaries. So even a GDP boom driven by capital returns translates into smaller tax revenue gains for the government than a boom driven by jobs and wages.
Meanwhile, a labour market disrupted by AI would almost certainly demand massive government spending to support displaced workers. Those two forces work against the fiscal math.
Why is the Treasury now struggling to borrow despite the world’s deepest capital markets?
For decades, foreign central banks reliably bought US Treasury bonds to build foreign currency reserves and manage exchange rates. That demand has evaporated. The Treasury now competes for private investment, and it is losing that competition to the very AI companies on which recovery hopes rest.
These superscalers, which the source identifies as Google, Meta, Microsoft, Oracle, SpaceX and Amazon, are borrowing heavily to fund datacenter buildout. A Stanford economist calculated that to break even on $1.43 trillion in AI infrastructure investment this year alone, datacenter owners would need revenue to grow 45% annually for seven years. By 2032, their AI revenue would need to equal roughly 9.2% of US GDP. A parallel analysis by economists Jared Bernstein and Ryan Cummings concluded the six superscalers need $13.1 trillion to $18.7 trillion in additional revenue over the next decade just to service their AI debt, equivalent to their total revenues over the last ten years.
This creates a vicious cycle. Rising interest rates, partly driven by AI companies’ own borrowing, push up the government’s cost of servicing federal debt, widening the deficit further. Foreign investors grow more cautious. The Treasury must offer higher yields to attract buyers. And if the AI companies miss their growth targets, the hit to private sector earnings could spook the bond market entirely.
The fiscal hole keeps growing while growth stays theoretical
Trump has since added another layer to the deficit problem: a promise of a $5,000 “dividend” to every adult if Republicans keep control of Congress in the midterms. That alone would cost roughly $1.7 trillion per year, with no funding mechanism identified.
The core truth is that no historical precedent, no model, and no reasonable growth scenario can reconcile the scale of current tax cuts with the size of the deficit and the direction of interest payments. The ai growth budget deficit trap is real: even if artificial intelligence supercharged the economy, its impact on the government’s finances would be muted unless the tax system shifts dramatically to capture a larger share of capital gains and corporate profits. Without that shift, even a productivity miracle will not prevent decades of either higher interest rates, inflation, or fiscal crisis.
For a deeper explanation of how government borrowing works and what drives Treasury yields, visit Thewealthora’s guide to how the federal deficit affects your investments.
Related coverage
- Anthropic AI model false information: why it matters
- Oil prices fall despite Middle East tension and US hurricane threat
- Nikkei falls on machinery stocks and Iran tensions
- Webull China ties expose U.S. investor data to Beijing
More stories like this on the Finance Wire.
Originally reported by The Guardian. Facts verified; analysis and wording are Thewealthora’s own.