AI: A Cure for the Deficit Narrative?

The “deficit narrative” has been prominent in financial media recently. This commentary will address the “deficit narrative,” potential outcomes, and how Artificial Intelligence might offer a solution.

The concern is that a bill winding through Congress will increase U.S. debt and deficit levels. Much of the concern comes from projections of a continuous rise in debt. The bill in Congress has been estimated to add around $5 trillion to the debt.

The U.S. has accumulated debt levels not seen since World War II, when federal debt exceeded 100% of GDP. The U.S. has generally run a negative deficit-to-GDP ratio since 1980. A surplus occurred briefly during the Clinton Administration, achieved by using Social Security funds to balance the budget.

While there is concern about the current 6% deficit-to-GDP ratio, it's not significantly above the average of -4%. The fiscal year 2024 deficit was $1.8 trillion, lower than during the COVID pandemic.

Government debt is the household’s asset. When the government issues debt, it enters the economy for purposes such as infrastructure and social welfare. That money is then credited to the bank accounts of households and corporations. Therefore, an increasing deficit results in increased economic activity, and vice versa. Demands for sharp deficit reductions may lead to an economic recession.

The deficit trend is worrisome because it continues during a period of economic growth that is slowing. Deficits typically decrease during economic prosperity due to higher tax revenues and decreased safety-net spending. Since the pandemic, deficits have decreased but are now increasing again due to economic weakness. The structural imbalance is putting upward pressure on interest rates, and central banks globally have stopped supporting debt markets.

If the “deficit becomes critical” due to higher interest payments, the Federal Reserve might reverse its monetary policy. The lack of interventions has allowed interest rates to rise. Currently, rates are trading at levels equating to economic growth and inflation. Interest rates seem high because they are returning to normal after pandemic interventions.

The consequence of that lack of intervention and the normalization of rates is the cost of servicing the national debt. In 2024, net interest spending reached $882 billion, surpassing expenditures on Medicare and defense. Due to the rise in rates, this figure has more than tripled since 2017, when it was $263 billion. If the Federal Reserve cuts rates and borrowing costs decline by 1%, the interest payment expense will decline by nearly $500 billion.

Interest payments are only one aspect; the other is revenue from economic growth. There are issues with Congressional Budget Office (CBO) projections. CBO projections are often biased, exclude data, and have structural flaws, which can mislead.

The most significant flaw is the dismissal of potential upside scenarios, such as demographic shifts or productivity surges due to technology. Conversely, it fails to adequately model downside risks like recessions or geopolitical shocks. This results in a misleading “middle path” that rarely reflects actual outcomes.

Previous high debt levels did not lead to economic devastation. Following WWII, strong economic growth helped shrink the debt ratio to about 25% of GDP by the 1970s. The 1980s military buildup and tax cuts increased debt. The 1990s tech boom and fiscal restraint decreased it. Events like 2008 and 2020 increased debt again. Currently, U.S. public debt is about 120% of GDP. While high debt isn’t new, the current trajectory stands out in scale and persistence.

There are reasons to hope the “deficit narrative” will improve. Following WWII, the US was the manufacturing epicenter. Global demand for US manufacturing increased economic growth rates, lowering debt-to-GDP ratios. The US is at the beginning of the next industrial revolution with Artificial Intelligence.

Construction growth is forecast to reaccelerate to 4% in 2026. The drivers of that growth will come from the buildout of Artificial Intelligence. If that activity only maintains the economy’s current growth rate, the debt-to-GDP ratio becomes more sustainable. This assumes interest rates do not fall, and spending remains at the current growth rates. However, if the growth rate increases to 4% annually, the debt-to-GDP rate will fall to roughly 100% by 2035.

Major corporations have already committed $1.8 trillion. These projects are either “shovel-ready” or near that stage. While $1.8 trillion may not seem like much, these are heavily infrastructure-intensive, which can significantly boost economic growth, directly impacting the debt-to-GDP ratio.

The U.S. GDP growth rate is projected at 1.8% annually through 2035 by the CBO, but infrastructure spending could push this higher. Every $1 billion in infrastructure investment creates 13,000 jobs and adds $3 billion to GDP over a decade. Therefore, if the U.S. invests $1.8 trillion in AI infrastructure by 2030—plausible given the $500 billion energy need, $300 billion for data centers, and $200 billion for chip production—GDP could rise by $5 trillion over 10 years, or roughly $300 billion annually.

Spending is expected to reach $6 trillion by 2030, equating to $18 trillion in economic growth. This spending will reduce the deficit trajectory. The current “deficit narrative” overlooks the economic improvements resulting from artificial intelligence’s buildout.

AI and Electricity Demand

The United States’ power grid has lacked sufficient investment to handle the increasing burdens of electricity demand in previous years. Adding electric vehicles, bitcoin mining, and artificial intelligence will overwhelm the current electricity supply in the U.S. Bitcoin mining demands an extreme amount of electricity.

AI energy demand is projected to surge from approximately $527.4 million in 2022 to a substantial $4,261.4 million by 2032.

AI Factories

The future will be the buildout of “AI Factories,” as we saw post-WWII. As more and more companies adopt Artificial Intelligence, every company competing in E-commerce, science, medicine, manufacturing, or service delivery will need to adopt AI, either directly or indirectly. The physical infrastructure of roads, buildings, power grid, housing, etc., must scale as AI factories come online. The U.S. has 2,700 data centers in 2024, but industry experts estimate a 50% increase is needed by 2030 to support AI growth. This buildout necessitates improved broadband infrastructure, as AI applications require low-latency, high-bandwidth networks. 5G coverage must expand to 90% of the U.S. population by 2028, up from 70% in 2024, to support AI data flows.

Investment Opportunities

The infrastructure requirements will be enormous to build these factories, including the utilities required for electricity provision. The hardware supply chain is critical. AI relies on specialized chips. The 2022 CHIPS and Science Act allocated $52 billion to boost domestic semiconductor production. The U.S. must double its chip manufacturing capacity by 2030 to reduce reliance on foreign supply chains and meet AI needs.

Conclusion

The infrastructure buildout for AI data factories can drive economic growth by creating jobs, stimulating industries, and enabling AI-driven productivity gains. Increasing growth only marginally would stabilize the current debt-to-GDP ratio. Boosting GDP growth to 2.3%- 3% annually would vastly improve outcomes. If interest rates drop by just 1%, this could reduce spending by $500 billion annually, helping to ease fiscal pressures.