CIO Insights

CIO Insights: The AI buildout faces a political test

19 November 2026

The Democratic sweep in the US midterm elections has cast doubts over the pace of the AI buildout, raising concerns about what a slowdown could mean for the broader economy. With AI investment a growing source of support for US growth, could a setback be enough to tip the economy into recession?

In this month’s CIO Insights, we examine what a slower AI buildout could mean for US growth and consider the implications for investors.

Key takeaways 

  • AI investment is an increasing source of support for US growth. Spending on data centres and equipment has supported economic activity, with further expansion expected in the years ahead. Some of that spending goes overseas or pays for imported equipment, so its direct contribution to US growth is smaller than headline figures suggest. Even so, a slower buildout would weigh on the wider economy.
  • Political headwinds could slow the AI buildout without derailing US growth. Tighter restrictions on data centres could weigh on spending, though projects already underway provide some near-term support. Our analysis suggests that even a substantial shortfall in planned AI investment could leave the US economy expanding, provided other growth drivers hold steady. The scale and timing of any disruption will depend on how any  restrictions develop.
  • Companies can adapt and capital can move, reinforcing the case for diversified global exposure. Constraints in some states may give companies an incentive to find new ways to build and power data centres, or invest in other states and countries, though these adjustments take time. That could sustain the broader AI buildout while moving some of its economic benefits elsewhere. For investors, diversified global equity portfolios offer exposure to opportunities across markets as companies adapt to these challenges and the locations benefiting from AI investment shift.
  • Look beyond political headlines and keep the longer-term investment outlook in focus. Past elections offer some perspective, but outcomes need to be considered alongside the economic and business conditions of the time. Following the latest midterms, that includes assessing what a slower AI buildout could mean for companies and investors. While markets could remain volatile in the near term, we remain constructive on AI’s potential to improve productivity across the economy.

(See our Glossary at the end for a breakdown of the terms used in this article.)

AI investment has increasingly supported US growth

The AI buildout isn’t just a story about the tech sector. As companies spend more on data centres, servers and other equipment, that investment has become an increasing source of support for the wider US economy. As Exhibit 1 below shows, its contribution to US growth began to pick up in early 2025 as the AI buildout accelerated.

Exhibit 1: AI investment’s contribution to US growth has risen in recent years

JPMorgan estimates that AI-related investment accounted for more than one-fifth of US economic growth in the year to June 2026. That’s equivalent to about half a percentage point of the economy’s 2.1% expansion over the period.¹ Spending on computers, related equipment, and software drove much of the increase in investment.

That contribution, though, is measured after accounting for imported hardware. Some of the spending on the AI buildout pays for equipment produced overseas, so looking at investment alone would overstate its direct contribution to US growth.

Estimates also depend on what counts as AI investment. Categories such as computers and software include spending unrelated to AI, making its precise contribution difficult to isolate. ING, for example, puts the contribution from technology investment at around one-third of US growth after adjusting for imports.²

While estimates can vary, they point to AI-related investment becoming a more meaningful source of support for the economy. The question is how much a slower buildout could weaken that support.

Political headwinds could slow AI investment without derailing US growth

The US midterm elections have brought local opposition to data centres into sharper focus. Communities have raised concerns about their impact on electricity grids, power bills, and water supplies, prompting calls for pauses on new projects and tighter requirements. State and local authorities control many of the approvals these projects need, giving newly elected officials considerable influence over the pace of the buildout.

Texas, Georgia and Ohio account for 44% of the US data-centre pipeline. Including Virginia, Pennsylvania, Illinois and Arizona, these seven states with Democratic governors account for roughly 70% of planned capacity, as Exhibit 2 shows.³

Exhibit 2: US states with Democratic governors account for a substantial share of planned data centre capacity

States like Texas, Georgia, and Ohio are seeing proposals to pause new development while stronger protections are put in place.⁴ Not every state is headed for a moratorium, however. Governors in Pennsylvania and Virginia, for example, haven’t indicated plans for one, but other requirements could still raise developers’ costs or delay projects.

To illustrate what disruptions to the buildout could mean for the US economy, we consider two hypothetical scenarios using these seven states’ roughly 70% share of planned capacity as a proxy for their share of US AI investment. Exhibit 3 shows how GDP growth could be affected if all or half of that spending were halted in either 2027 or 2028.

Exhibit 3: US growth could prove resilient even if AI investment slows sharply

Holding other drivers of growth constant, we estimate that AI investment coming in 35% below those projections could slow US real GDP growth from a 2.2% baseline to around 1.8% in either year. In a more adverse scenario, with a much steeper shortfall of 70%, growth could fall to around 1.4% in 2027 or 1.3% in 2028.

That’s because these spending cuts could have a more limited impact on GDP growth than their size might suggest. Drawing on Federal Reserve research on data centre spending, we assume around 44% of AI investment goes to US-produced goods and services, with the rest paying for imported equipment.⁵ Part of any spending reduction would thus fall on equipment produced overseas, cushioning the direct impact on the US economy.

The impact would also depend on how far projects have progressed. Any restrictions on new development would be more likely to affect future projects, while construction that’s already underway could continue to support investment.

About 75% of US data-centre capacity scheduled to come online in 2027 is already under construction, making those projects less vulnerable to new restrictions. For 2028, though, projects accounting for 39% of the 96 gigawatts (GW) of capacity scheduled to come online could face delays if restrictions were to come in place.⁶

Our analysis therefore suggests that slower AI investment would shift US growth down a gear, but could still leave the economy in expansion, even in our more adverse scenario. 

AI investment can find new paths, making the case for diversified exposure 

The concentration of planned data-centre capacity in a handful of US states leaves the buildout exposed to local restrictions. Companies can direct spending towards other locations, but developing new sites – from securing the land to power connections and permits – takes years. BloombergNEF estimates that hyperscalers take around five years to bring a data centre from development to operation, limiting how quickly new projects could replace stalled capacity.⁷

But as long as demand remains strong, companies have an incentive to find alternatives. Directing future spending towards other US states could preserve some of the benefits for the US economy. Outside the US, Morgan Stanley estimates that Asia could absorb around 15 gigawatts of additional data-centre demand from US-linked projects as power constraints could push investment overseas.⁸

Expanding abroad could sustain the global AI buildout while shifting some of its economic benefits to other countries. Other markets face their own power and permitting constraints, however. Location also matters more for some uses of AI than others: model training can take place farther from users, while services that depend on rapid responses may need facilities closer to customers.

Companies can also change how they build and operate their facilities. Microsoft’s newer cooling designs, for example, circulate water through a closed loop, avoiding continual losses through evaporation.⁹ Advances like these could ease some of the pressure on local resources and make further expansion more feasible.

For investors, this reinforces the case for diversified global equity exposure. Opportunities extend across companies developing AI, their suppliers, and businesses adopting it. As companies adapt and shift, investing across sectors and regions reduces reliance on any single part of the buildout. (Read more in our CIO Insights: Can AI afford its ambitions?)

Take the long view on politics, AI, and markets 

History offers some perspective, though few midterms closely resemble this one. Across the 19 midterm elections from 1950 to 2022, Democrats regained control of both chambers of Congress on just three occasions: 1954, 1986 and 2006. All of those left a Democratic Congress facing a Republican president. 

As Exhibit 4 shows, the S&P 500 rose over the following 12 months in all three cases, but the gains varied widely: 33.2%, 1.1% and 6.7%, respectively. That variability in returns is a reminder that similar political outcomes can play out against very different economic and market conditions. While historical patterns can be a useful starting point for investors, what comes next depends on the circumstances of the time.

Exhibit 4: Returns varied widely after Democrats regained control of Congress

In the context of AI, taking a more nuanced, longer-term perspective also means recognising that a slower buildout need not be a bad thing. A more measured pace of investment could give hyperscalers more room to assess where their spending will deliver the greatest value. It could also give investors time to consider opportunities without feeling pressured to chase the rally.

At StashAway, we remain constructive on AI’s longer-term potential to improve productivity across the economy and transform how we work and live. But as we’ve shared recently, uncertainty over the AI buildout could bring greater market volatility in the near term. Maintaining diversified exposure and investing steadily can help investors participate in the technology’s potential while keeping their long-term goals in focus.

Authors

Stephanie Leung, Chief Investment Officer

Stephanie and her team oversee the full spectrum of investment products and portfolios offered at StashAway. She brings more than two decades of investment expertise across multiple asset classes. Prior to joining StashAway in 2020, she managed investment portfolios at institutions such as Goldman Sachs and multi-billion dollar family offices in the region.

Justin Jimenez, Head of Macro and Investment Research

Justin brings 15 years of experience in economic and investment research to StashAway, where he contributes to shaping the investment office’s views on the global economy and financial markets. Before joining StashAway in 2022, he was an economist at Bloomberg, and holds degrees in international economics and finance from Columbia and UCLA.

Glossary

US midterms

Elections held halfway through a US president’s four-year term. Voters choose members of the House of Representatives, about a third of the Senate, and officials at the state level, including governors in most states.

Data centre

A facility housing the computers and networking equipment used to store and process data, run applications, and train and operate AI models.

Data-centre pipeline

Projects planned or under development, including those already under construction. Pipeline capacity excludes facilities already operating and may not all be completed.

Moratorium

A temporary pause on an activity, such as approving or building new data centres, often while authorities review rules or introduce new requirements.

Hyperscaler

A large technology company operating extensive data-centre infrastructure to support its own services or provide cloud computing to customers. Examples include Amazon, Microsoft, Alphabet, and Meta.

Model training

The process of teaching an AI model to recognise patterns using large amounts of data, enabling it to perform tasks such as generating text or analysing information.

References

  1. Aliaga, S. (2026). How much of the AI boom is adding to U.S. growth? J.P. Morgan Asset Management. Retrieved from: https://am.jpmorgan.com/us/en/asset-management/institutional/insights/market-insights/market-updates/on-the-minds-of-investors/how-much-of-the-ai-boom-is-adding-to-us-growth/
  2. Knightley, J. (2026). How much is AI contributing to US economic growth? ING THINK. Retrieved from: https://think.ing.com/opinions/how-much-is-ai-contributing-to-us-economic-growth/
  3. BloombergNEF. (2026). US Election’s Blue Wave Hands Energy Reins to States. Retrieved from: https://www.bnef.com/analyst-reactions/tnzj1ckjh6v500
  4. Robins-Early, N. (2026) Several Democrats renew calls for datacenter pauses after blue wave. The Guardian. Retrieved from: https://www.theguardian.com/us-news/2026/nov/06/democrats-renew-calls-for-datacenter-pauses
  5. Brandsaas, E. E., Garcia, D., Kurtzman, R., Nichols, J., & Zytek, A. (2025). Estimating Aggregate Data Center Investment with Project-level Data. Finance and Economics Discussion Series 2025-109, Federal Reserve Board. Retrieved from: https://www.federalreserve.gov/econres/feds/files/2025109pap.pdf
  6. Cembalest, M. (2026). Now You Can See Everything: on the midterms, AI, markets, the Theranos film and corporate fraud. Eye on the Market, J.P. Morgan Asset & Wealth Management, p. 1. Retrieved from: https://assets.jpmprivatebank.com/content/dam/jpm-pb-aem/global/en/documents/eotm/now-you-can-see-everything.pdf
  7. BloombergNEF. (2026). Six Things to Know About BNEF’s New US Data Center Capacity Outlook. Retrieved from: https://about.bnef.com/insights/data-centers/six-things-to-know-about-bnefs-new-us-data-center-capacity-outlook/
  8. Maheshwari, M. (2026). From AI to Air Conditioning: The Power Boom in Asia. Thoughts on the Market, Morgan Stanley. Retrieved from: https://www.morganstanley.com/insights/podcasts/thoughts-on-the-market/asia-growing-power-demands-mayank-maheshwari
  9. Priest, J. (2026). Inside Microsoft’s two-decade push to cut water intensity while scaling for growth. Microsoft. Retrieved from: https://blogs.microsoft.com/blog/2026/06/24/inside-microsofts-two-decade-push-to-cut-water-intensity-while-scaling-for-growth/

Share this

Keep reading

You may also be interested in:

View all insights