Three-line summary
- Depending on the outcome of the U.S. midterm elections, the odds rise for tougher rules on data centers covering power grid costs, tax treatment and permitting
- If Democrats expand their influence in Congress, AI infrastructure investment costs and pressure on Big Tech cash flow could increase
- After chips, power and funding, the question is whether "politics" emerges as the newest bottleneck for the AI industry
The U.S. midterm elections are about 10 days away. The outcome is expected to affect not only American politics but also the economy, society and foreign policy. The artificial intelligence industry is unlikely to be an exception. In particular, the election results could determine whether regulation of data centers is tightened. That means the midterms could become a new variable for AI investment. In this article, we look at the midterm landscape that has emerged as a variable for the AI boom.
Democrats Favored in the House, Senate a Toss-Up
In the current U.S. Senate, Republicans hold 53 of the 100 seats, making them the majority. Democrats hold 45 seats and independents two. Republicans also control the House. Of the 435 seats, Republicans hold 218, Democrats 214, with one independent and two vacancies, leaving the two parties narrowly divided.
So what happens after the midterms? U.S. media outlets see a strong chance that Democrats will recapture the House majority. The Senate, by contrast, is widely expected to be a tight contest.
How would such a landscape shape U.S. legislation and administration? According to a recent Reuters report, if Democrats take back only the House — meaning Republicans keep their Senate majority — the party's powers of investigation, hearings and subpoena would be significantly strengthened. Still, Republicans would control the Senate, and President Donald Trump could exercise his veto. Reuters noted that even if Democrats pass bills in the House, Senate Republicans can block them, and Trump can still rely on executive orders.
If Democrats also take the Senate, they are expected to push legislation far more aggressively. That would be all the more likely if they win big enough to secure larger majorities in both chambers.
Democrats Push to Make Data Centers Pay
This landscape will also affect data center regulation. A prime example is the Ratepayer Protection Act, or RPA, which would have state regulators review standards for recovering from data centers the added cost of expanding generation, transmission and distribution facilities when large data centers make such expansion necessary. The bill passed the House last month by an overwhelming 417-3 vote, indicating that most Democratic and Republican lawmakers supported it.
In the Senate vote last month, however, the bill fell short, with 57 in favor and 43 against — short of the 60 votes needed to advance. A similar measure, the GRID Savings Act introduced by Democratic Sen. Martin Heinrich, would actually require data centers to bear the costs, and the main objection to the RPA was that it goes no further than a recommendation. Democrats argue the measures are not meant to restrain the AI industry but are consumer protections to keep grid costs tied to data center expansion from being passed on to households and small business owners.
That suggests Democrats take a harder line on shifting the cost of data center expansion onto ordinary electricity ratepayers. If they win the midterms, that stance looks likely to grow stronger. Public opinion among Democratic voters has also played a role. In a Reuters/Ipsos poll this month, 69% of registered Democratic voters said they supported limits on data center construction. Among registered Republican voters, 43% said they supported such limits.
Another Variable Squeezing Big Tech Cash Flow
This is not simply about tighter regulation. It means grid expansion costs would be added to the astronomical AI infrastructure spending that U.S. data center operators, including hyperscalers and neoclouds, are pouring out. According to the London Stock Exchange Group, free cash flow at the five companies — Microsoft, Alphabet, Amazon, Meta and Oracle — is projected to swing to a deficit of $2.87 billion next year, as their capital spending soars to $945.55 billion over the same period.
Bloomberg also reports that Democrats in both chambers of Congress have introduced about six bills that would scale back tax breaks for data centers or impose taxes on AI use and excess profits. At the heart of the AI profitability debate now rattling markets is the concern that AI companies, facing heavy cash burdens from massive infrastructure spending, may fail to generate adequate returns. If Democrats sharply expand their influence in Congress in the midterms, that could become a new policy risk driving up Big Tech's AI infrastructure costs. Even if the bills are never enacted, the mere possibility of tougher tax treatment, grid cost rules and permitting conditions could alter expected returns on data center investment and financing terms. As the AI expansion accelerates, Big Tech — which is betting everything on infrastructure investment — may run into another bottleneck in politics and regulation.
Of course, election results cannot be known until the votes are counted. U.S. media outlets see a low chance that Republicans hold majorities in both chambers, but they do not rule it out entirely. Ultimately, whether politics becomes another bottleneck for the AI industry depends on the outcome of the midterms.
Related "Petro-Electro" articles
1. Are data centers only guzzlers of water and electricity?
2. The "AI electricity bill" is coming. Who pays?
3. The AI profitability debate: even the power guzzler faces doubts
From petroleum to electricity, energy is a key keyword for understanding the economy, industry, international affairs and the response to climate change. Subscribe to the reporter or to the "Cho Yang-joon's Petro-Electro" series for economic insights seen through energy.