AI, data centers and US growth

At the beginning of the current year, the US economy enjoyed numerous tailwinds. The Federal Reserve appeared poised to cut rates further on any sign of employment weakness. The One Big Beautiful Bill offered stimulus via tax refunds and capital spending incentives. Deregulation was in the works, especially for energy and banking. The tariffs uncertainty that characterized 2025 was increasingly in the rearview mirror. Furthermore, AI presented optimism for future economic growth, both in its build-out and implementation. But the war in Iran introduced uncertainty to most of this bullish backdrop.

The bull case for 2026 and beyond is increasingly dependent on the promise of AI, even as the debate about pros and cons rages in real time. I am bullish on AI, but it is worth considering the potential negatives as we enter this brave new world.

If we build it, will it be the last thing we do?

It is probably the case that any truly revolutionary change comes with drawbacks. But few advances have had the perception problem AI has. Widespread adoption of AI threatens growth if whole sectors of the economy are rendered redundant, slow-adopting businesses fail and demand for labor falls. The US is a consumer-driven economy. If layoffs are the main impact of AI, the boost to corporate bottom lines and economic growth will be fleeting, to say the least.     

Then there is the environmental side. Data centers, the heart of the AI ecosystem, are extremely energy intensive. The typical AI-optimized data center uses as much electricity as 100,000 households. This drives up power costs and adds stress to already-overtaxed energy grids. Data centers also use huge amounts of water, potentially creating environmental issues and competing with residential water demand.

How long will it be before protests against data centers become widespread? We have already seen some resistance become codified, with Maine recently passing a law banning their construction. As with fracking—which has turned out to be a vital competitive advantage for the US as a whole—this sometimes may just be NIMBY-ism. In other cases, it is prudence. You cannot put a data center just anywhere. In addition to water and energy, they also require large tracts of land to be feasible.

The initial promise of a boost to the local economy from a new data center is tantalizing to municipalities. The build-out itself provides many high-paying construction jobs. There is a catch, however: once completed, a data center offers few permanent jobs. It is more like a highway, say, than a factory. The multiplier effect will need to be considered more to boost productivity.

The bull case

These are all worthy concerns, but they seem short-sighted and take too narrow a perspective. It is inevitable that creative destruction will bring disruption, but the history of prior technological advances shows that an economy ultimately grows as activity broadens.

Some of the more pessimistic projections already appear to have been misplaced, and areas thought to be most vulnerable are actually doing well. For example, software job openings (per data from Indeed) are increasing and stand at their highest level in nearly two years. There is already a new term — “AI-washing” — for companies using AI as the excuse for trimming staff. The reductions give the appearance of improved efficiency, when in reality they are due to overstaffing or simply the normal course of business.

Importantly, some studies show that, contrary to expectations, AI adopters are increasing the intensity of their work. Rather than cutting back on overall effort, the capacity that AI frees up essentially gets repurposed into other constructive activity. AI promises to improve productivity, drive down costs and increase demand for products and services across the economy. By efficiently executing low-end tasks, it can promote high-quality job creation and increase the capacity for innovation and research. At the end of the day, the potential merits of AI are too enticing for the logistical and resource constraints not to be overcome. Also, bear in mind that the US government has identified data center expansion as critical to national security and overall global competitiveness.

At the individual job, company and even industry level, there will undoubtedly be winners and losers as more efficient adopters outperform the competition. But the fear that this new technology will broadly lead to negative outcomes across the economy as a whole seems misplaced. Rather, a better perspective may come through the lens of Adam Smith’s “Invisible Hand.” People make decisions in their self-interest, and that competition enhances behavior. The result can be an efficient allocation of resources benefiting society. With change comes fear, but obsessing about the near-term uncertainty AI brings is a case of not seeing the forest for the trees.