Big Tech’s Data Center Assumptions Are Unraveling—At Home and Abroad
April 15, 2026
The state of Maine and the United Arab Emirates do not often appear in the same sentence. But over the last month both—in different ways—have undercut some crucial assumptions Big Tech companies have made about the long-term viability of their AI infrastructure rollout.
On April 6, the Maine House of Representatives passed Bill LD307, which would put a moratorium on the construction of large data centers in the state until November 2027. It also created a Data Center Coordination Council, which would evaluate new proposals and facilitate planning considerations. The bill is expected to pass the state Senate and be signed into law by Governor Janet Mills (D).
At least eleven other states are considering similar moratoriums, and in late March Sen. Bernie Sanders (I-VT) and Representative Ocasio-Cortez (D-NY) introduced legislation for a federal moratorium on AI data centers. While it’s unlikely that either the federal or all the state moratoriums will pass, they are nonetheless reflective of abroad domestic sentiment against AI, especially how data centers impact the environment, home energy costs and local job opportunities.
Meanwhile in the Middle East, the war between Iran and the United States has given Big Tech companies a rude awakening about the strategic vulnerability of their multi-billion data center investments in the region.
On March 7, Iranian drones damaged three Amazon Web Services (AWS) data centers located in the UAE and Bahrain. On April 2, Iran claimed to have damaged an Oracle data center in the UAE—although the UAE denied the report. On April 7, the Iranian Revolutionary Guard Corps (IRGC) released a video in which they promised to target “information and communications technology of the Zionist regime” should the US target Iranian power plants. The video then showed a night vision view of Stargate, a $30 billion ongoing data center investment by OpenAI, located in the UAE.
What links these disparate incidents is a set of mistaken assumptions made by Big Tech companies that they could follow a tried-and-true manual of moving fast and breaking things with minimum consequences.
Within the US, there was an assumption that the political economy that made data centers welcome (such as the Trump Administration’s July 2025 Executive Order fast-tracking Federal Permitting for Data Centers) would continue to hold. The infrastructure would be treated as the“backbone of the 21st century’s economy,” and states would offer significant tax exemptions in return for future generation of economic activity and jobs. This would help to position states as national hubs for a burgeoning AI industry.
Within the Gulf, it was assumed that the region’s combination of cheap energy, massive sovereign capital, a deregulated business environment and geopolitical neutrality could help countries like Saudi Arabia and the UAE shift towards a post-oil future. Significant capital has flowed into both bets. Saudi Arabia, for instance, has moved resources from its $500 billion proposed NEOM megacity towards AI infrastructure, while the UAE launched a $100 billion investment vehicle, MGX, geared explicitly towards AI investment. Taken together, these represent hundreds of billions of dollars staked on assumptions that are now looking shaky on both fronts.
Gulf states might still have the deregulatory environment but can no longer claim geopolitical stability. The United States has that (domestic) geopolitical stability, but the examples above show an appetite for increasing regulation when it comes to data center buildouts.
It is clear then that Big Tech companies need to read the room better and adjust their approach when it comes to data centers. But spending tens of millions on lobbying to counter AI regulation (like both Meta and OpenAI have done) isn’t the answer. Companies instead need to advocate for converting their voluntary commitments into minimum enforceable standards that cover electricity price increases, water use, job impact and other community concerns. These standards need to apply universally, not just to willing participants.
Microsoft, Anthropic and OpenAI have all released initiatives that make sustainability and equity promises (the first point of Microsoft’s Community-First AI Infrastructure Plan, for instance, is to ensure that data centers don’t increase electricity prices). But these commitments remain voluntary and selective (neither AWS nor Google have yet made equivalent pledges) and impossible for local communities to effectively verify. That gap is precisely what enforceable federal legislation would close.
Both the domestic US backlash and geopolitical challenges that data centers are now facing in the Gulf show the mistaken assumption of US Big Tech companies that accountability can be deferred indefinitely. But now the results of deferring that accountability have been compounded, leaving AI investors in a bind. They now have the opportunity to instead proactively engage in democratic processes and advocate in good faith for a federal framework that creates genuine accountability, including measures such as robust AI auditing standards and transparency mandates for data center electricity use. In doing so Big Tech companies will build more political legitimacy and, with it, infrastructure that is considerably harder to undermine.
Technology & DemocracyRelated
See all
