US Midterm Election Results Raise Concerns for AI Infrastructure Expansion

“Data centers in space looking pretty appealing right now,” Elon Musk recently posted on X as the results of the US midterm elections became clear. The comment was meant as a joke. But it also highlighted a growing problem for technology companies. Building the infrastructure needed to power artificial intelligence (AI) is becoming increasingly difficult in several parts of the US. According to multiple media reports, Musk spent $500 million (about Rs 4,700 crore) backing Republican candidates. His post came after data center opponents emerged as the biggest winners of the Democrats’ landslide victories in the US midterm elections, pointing to a shifting public mood on AI. Besides gaining control of both the House of Representatives and the Senate, Democrats also won seats in traditionally Republican states such as South Carolina, Texas and Kansas. Several Democratic candidates made restrictions on new data centre projects a key part of their campaigns. This could have important consequences for the AI industry. Data centres are essential for running AI applications, but they require enormous amounts of electricity and water. Growing concerns about their impact on local communities have led to calls for stricter rules. The election results could therefore change where technology companies build new data centres, how much these projects cost and how quickly they can be completed. However, the immediate reaction from the stock market was unexpected. When markets opened after the results, some of the biggest technology companies saw their shares rise sharply. Investors appeared to welcome the possibility of a slower expansion of AI infrastructure. A slowdown could allow large technology companies to reduce their massive spending on new facilities and improve their cash flow. However, this does not necessarily mean that investment in AI will fall. Companies could continue spending on AI chips, software and computing services even if building new data centres becomes more difficult. The larger question is whether the election results will change the pace, cost and location of America’s AI investments. The biggest concern for the industry comes from three states: Texas, Georgia and Ohio. Democrats won the governors’ elections in all three states. These states account for about one-third of the planned data centre capacity in the US, according to ICICI Bank Global Markets report. In its November 4 report, ICICI described the results as “a very consequential development”. The three states have a combined data centre capacity of 156 gigawatts, according to ICICI. Of this, around 141 gigawatts is planned and approximately 15 gigawatts is already operational. A gigawatt is a unit used to measure power. It is roughly equivalent to the electricity demand of a large city. The difference between existing and planned capacity is important. Projects that are already operational or under construction are less likely to face immediate disruption. However, projects that are still in the planning stage could face delays or cancellation if governments introduce new restrictions. ICICI’s assessment was that “now a significantly larger share of planned data center capacity is exposed to political risks than previously thought.” Goldman Sachs economists, quoted by NBC News, also warned that Democratic victories at the state level “are likely to lead to short-term data center moratoria, potentially affecting about 45% of planned US data center capacity expansion.” Goldman also identified Michigan as a state where temporary restrictions could emerge. This means a significant portion of America’s planned data centre expansion could face political and regulatory uncertainty. Gina Hinojosa defeated Republican Governor Greg Abbott in Texas. During her campaign, she called for an immediate pause on new data centre projects. “We need an immediate moratorium on data centers to last until the legislature can meet to put in place real protections so Texans can have a seat at the table to decide if and how data centers come into their communities,” she said, as quoted by AFP. ICICI said Hinojosa also wants to end tax benefits given to data centre developers. Texas is particularly important because it accounts for 22% of the country’s data centre capacity, according to ICICI. AFP, citing Bloom Energy, reported that the state is expected to account for nearly 30% of national electricity demand for AI operations by 2028. Any restrictions on new projects in Texas could therefore affect a large part of America’s planned AI infrastructure. In Georgia, Keisha Lance Bottoms won the governor’s election, ending more than two decades of Republican control, according to Nomura. ICICI said Bottoms has supported a statewide pause on new data centre projects while the government studies their impact on electricity bills, water supplies and local communities. Georgia accounts for around 5% of the country’s data centre capacity, according to ICICI. The proposed pause does not necessarily mean a permanent ban. However, it could delay new projects while the government examines whether the benefits of hosting these facilities outweigh the costs for residents. Ohio’s incoming governor, Amy Acton, has proposed a different approach. Instead of a complete ban, ICICI described her proposal as a conditional moratorium. This means developers may be allowed to build data centres, but only if they meet certain conditions. These include requirements covering electricity costs, environmental protection, transparency, employment and benefits for local communities. Such conditions could increase construction costs and make the approval process longer. Even if a state does not completely stop new projects, stricter rules can make it more difficult and expensive for companies to build them. The election results come at a time when investors are increasingly concerned about how much money large technology companies are spending on AI. Building data centres, buying advanced computer chips and setting up the necessary power infrastructure require huge investments. Companies such as Amazon , Microsoft , Alphabet and Meta have been spending heavily to expand their AI capabilities. However, investors have started questioning whether these companies will be able to generate enough revenue from AI to justify such large investments. The possibility of government restrictions has now created another way for these companies to slow down their spending. If governments delay new data centre approvals, technology companies may not have to immediately commit money to projects that cannot be completed on time. This could improve their cash flow, which refers to the money coming into and going out of a company. The stock market reaction showed that investors were looking at this possibility. According to Bloomberg, Amazon shares rose as much as 8.3% on Wednesday. Microsoft gained 7%, Alphabet jumped 5.9% and Meta advanced 3.6%. Apple, which has taken a more cautious approach to large-scale AI infrastructure spending, climbed more than 6%. Five companies were on track to add nearly $1 trillion to their combined market value in a single trading session, Bloomberg reported. Reuters also reported gains of around 6% each for Amazon and Microsoft. Robert Pavlik, senior portfolio manager at Dakota Wealth Management, told Bloomberg, “The market is acting like this represents the end of the AI buildout.” He added, “If that means the hyperscalers aren’t spending ridiculous amounts of money to build out compute, that means they could become cash-flow positive again.” Hyperscalers are large technology companies that operate huge data centres and provide computing services to businesses and consumers. Pavlik explained that these companies were facing criticism from investors for spending too much money. At the same time, cutting their investments could make investors question whether they still believed in the future of AI. “Hyperscalers had been stuck between a rock and a hard place,” Pavlik said, “where they were getting punished for spending, but pulling back would have also raised questions.” Government restrictions could give these companies a reason to slow down their investments without making it appear that they had lost confidence in AI. However, a rise in share prices does not mean these companies have already decided to cut their AI budgets. It reflects investors’ expectations about how the political changes could affect future spending. Though shares of the biggest technology companies rose, companies directly involved in building and supplying data centres saw their shares fall. Reuters reported that Applied Digital and CoreWeave fell 8.8% and 9.5%, respectively. Cipher Digital dropped 21%. Companies in the power sector also faced losses. GE Vernova declined 11.5%. Constellation Energy and Vistra, both of which have significant exposure to data centre projects in Texas, lost 9.2% and 15.4%, respectively. The difference in market performance comes down to the kind of business these companies operate. Large technology companies have already built substantial infrastructure. They also have stronger financial positions and may be better equipped to manage delays in new projects. However, companies that supply equipment, provide computing capacity or depend on new data centre construction could face immediate problems if projects are postponed or cancelled. Companies that borrow large amounts of money to finance new facilities could be particularly vulnerable. Mark Malek, chief investment officer at Siebert Financial, explained this difference while speaking to Reuters. “The AI buildout will become more expensive, not disappear.” He added, “The most vulnerable AI companies are the borrowers, not necessarily the biggest spenders.” In other words, companies with strong balance sheets and existing data centre networks may be able to manage regulatory delays. Businesses that depend heavily on borrowed money could struggle if costs rise or projects take longer to complete. Patrick Munnelly, market strategist at Tickmill Group, made a similar point to AFP. “None of this necessarily breaks the long-term AI theme, but it can raise the hurdle rate for investment,” he said. This suggests that the election results could affect individual companies differently rather than bring the entire AI investment cycle to a halt. One possible response to stricter rules is that technology companies could move their investments to states that are more welcoming to data centre projects. This could change the geographical distribution of AI infrastructure in the US. For instance, if Texas introduces restrictions on new projects, companies could consider building facilities in states where approvals are easier and electricity is more readily available. Sarah Bianchi of Evercore warned that “2027 is shaping up to be a turbulent year as developers will be forced to adapt to a shifting policy landscape.” However, she said the overall impact “will hinge on the details,” including whether developers can move projects to states with fewer restrictions. Emkay Global Financial Services also indicated this possibility. The Indian brokerage said state-level restrictions “will simply cause firms to move data centers to states which are more welcoming.” This means the next phase of the AI boom may depend less on whether companies continue building data centres and more on where they choose to build them. However, shifting projects is not always easy. Companies must consider the availability of electricity, land, water, power connections and other infrastructure before selecting a location. A state with fewer restrictions may not necessarily have all the resources needed to support a large data centre. Although the election results have raised concerns about future projects, the immediate impact could be limited.Nomura noted that much of the data centre capacity expected to become operational in 2027 is already under construction. “Much of the data center capacity expected to come online in 2027 is already under construction, limiting the immediate impact of restrictions on new permits,” the bank wrote. This means restrictions on new construction permits may not have a major effect on projects that are already being built. The bigger risk could emerge in 2028. A larger share of the capacity expected to become operational that year is still in the planning stage. These projects are more exposed to changes in government policy and could face delays if new restrictions are introduced. Nomura also cautioned against treating the Texas election result as an immediate threat to every data centre project in the state. Abbott had already introduced a temporary permitting freeze in September. Hinojosa’s proposed restrictions are also expected to exclude projects that are already under construction. According to Nomura, her concerns are largely centred on tax benefits and the use of local resources rather than opposition to AI itself. Goldman Sachs expects any restrictions to be relatively short-lived. Its economists wrote, “we expect that the issue will fade somewhat in importance in the ensuing months and that these moratoria will not be long-lived if they are enacted.” Morgan Stanley Research also said it expects “the broader AI capital expenditure (capex) cycle to remain intact.” Capital expenditure refers to the money companies spend on long-term assets such as buildings, equipment and infrastructure. Altimeter Capital’s Brad Gerstner told MarketWatch, as relayed by Fox Business, that fears of a halt to data centre construction were exaggerated. He estimated that 10 to 15 gigawatts of expected AI capacity could be affected. These views suggest that the election could delay some projects without necessarily changing the long-term investment plans of the technology industry. The change in control of Congress has raised questions about whether Democrats could introduce tougher national rules for AI infrastructure. However, passing new laws at the federal level may be difficult. President Donald Trump has the power to veto legislation passed by Congress. Democrats would need a two-thirds majority in both the House and the Senate to override his veto. Analysts have different views on whether this is possible. ICICI said securing a veto-proof majority “is unlikely.” Emkay, however, claimed that the possibility of an override is “materially higher now”, pointing to 10 to 12 vulnerable Republican senators who will face re-election in 2028 and could sometimes support Democratic positions. Even so, Emkay noted that any new legislation restricting AI would still need Trump’s signature unless Congress could override his veto. At the federal level, the election outcome “is less likely to change the outlook for data center and AI regulation, which would still need bipartisan support.” This makes state governments particularly important. UBS analysts argued that “The composition of state governments, including the result in Texas, may prove more consequential for individual data center projects than the change in congressional control.” Morgan Stanley also indicated the role of state and local authorities. Decisions on connecting data centres to electricity grids, setting electricity rates for large users, selecting locations and providing tax benefits are largely influenced by state regulators, state governments and local authorities. As a result, some states may continue to attract data centre investments, while others introduce stricter rules. Morgan Stanley described this as “a more fragmented environment”. For technology companies, this means that their investment decisions will increasingly depend on the policies of individual states. The impact of the US election results could extend beyond the country’s borders. The US is one of the biggest markets for AI infrastructure. Any slowdown in its data centre expansion could affect companies that supply computer chips, equipment and other technology products. However, analysts do not expect the election results to immediately disrupt the wider AI investment cycle. Emkay described the election as “a sentiment shock for AI (led by local anti-data center sentiments), not a capex shock.” The brokerage expects growth in AI-related capital expenditure to slow in 2027. However, it believes this will mainly happen because the investment base has become so large, rather than because of political restrictions. It therefore expects the AI trade and the semiconductor supply chain across emerging Asia to “stay supported.” This suggests that Asian companies supplying chips and other technology products to the AI industry could continue to benefit from demand, even if some US data centre projects face delays. Emkay’s bigger concern is a sharp tightening of credit conditions. If banks and investors become less willing to lend money, AI companies could find it harder to finance their projects. A slowdown in AI spending could also have wider economic consequences. Technology companies may need to borrow less money to fund new projects. This could reduce concerns that heavy borrowing by technology companies will compete with the US government for funds. Emkay said this could ease some pressure on borrowing costs. However, it does not expect either political party to take meaningful steps to address the US fiscal deficit. BlackRock Investment Institute also stated the relationship between AI investment and financial markets. It said changes in expectations around AI spending could “ease or intensify competition for capital”, influencing bond yields at a time when heavy government borrowing is already keeping them elevated. The US midterm election results have introduced a new challenge for companies planning to expand their AI infrastructure. New restrictions could make it more expensive and difficult to build data centres in some states. Projects could face longer approval processes, higher electricity costs and stricter environmental requirements. However, this does not necessarily mean technology companies will stop investing in AI. Large companies with strong financial positions may continue expanding their AI businesses while changing where and how they build their infrastructure. Smaller companies that depend on borrowed money and new construction projects could face greater pressure. The immediate market reaction also shows that investors are looking at the possibility of lower spending by Big Tech and better cash flows. The bigger impact may become visible in 2028, when more planned data centre projects could face regulatory uncertainty. The election results point towards a slower and more complicated expansion of AI infrastructure rather than the end of the AI boom. The future of AI investment in the US may depend not just on how much companies want to spend, but also on which states are willing to accommodate that spending.