The 'Big Three' AI giants have jointly called for a slowdown in AI development—does this signal the arrival of the technological singularity, or is it simply a sign of capital shortage in the market? How should we forecast the future of the AI economy?

Neither favorable timing, nor advantageous geography, nor popular support currently supports continued high-speed advancement by U.S. AI giants—so a moderate deceleration is the right choice. Of course, this means only a measured slowdown—not slamming on the brakes. 1. Timing, geography, and popular support all oppose runaway AI development. Timing—in the U.S. context—refers to financial markets. Funding for U.S. AI infrastructure has long shifted away from corporate internal cash reserves toward debt financing, equity issuance, equipment leasing, project finance, and supplier guarantees—with debt increasingly dominating. As of Q1 2026, the five largest hyperscale cloud providers plus NVIDIA have issued approximately $320 billion in debt. Goldman Sachs forecasts that debt issuance by hyperscale cloud firms and chipmakers will rise roughly 40% in 2027 compared to 2026. Meanwhile, U.S. Treasury sales pressure is also mounting, creating direct competition between AI corporate bonds and U.S. Treasuries. Compounded by inflationary pressures stemming from Trump’s provocations in Iran, maintaining U.S. financial stability now requires slowing down AI bond sales. Geography refers to AI compute center construction. In the U.S., such construction is increasingly constrained by site availability, electricity supply, water access, and other foundational infrastructure. In Q1 2026 alone, at least 75 data center projects—valued collectively at around $130 billion—were suspended or delayed across the United States. Popular support—or lack thereof—refers to public and legislative sentiment. Due to inadequate prior planning for AI compute centers, local communities have found themselves competing with these facilities for water and electricity, triggering widespread resident opposition. Media narratives amplifying fears about AI displacing ordinary workers have further fueled backlash. Restricting unregulated AI development has now become bipartisan consensus: Democrats consistently demand stronger oversight, while many Republicans—including prominent figures—are shifting toward supporting AI constraints. For instance, Greg Abbott, Republican Governor of Texas, issued an executive order in August suspending approvals for new data centers until state agencies complete assessments of their potential strain on public utilities. Similarly, Florida—the so-called 'stronghold of the know-it-alls'—has enacted 24 prohibitions targeting AI compute center construction. Thus, even before the 'Big Three' jointly urged global restraint on frontier AI model development, the U.S. AI industry had already lost momentum.

AI compute centers are competing for electricity and water; U.S. AI data centers face nationwide resistance.

Shapiro also announced he would remove all data center projects from Pennsylvania’s fast-track permitting program, rendering them ineligible for those expedited measures going forward. He further stated that, under his new directive, data center projects would no longer be permitted to use confidentiality agreements. In his speech, Shapiro noted that since last year’s announcement of two large Amazon-backed data center projects in the state—which he had enthusiastically promoted—Pennsylvania had been ‘flooded with speculative data center proposals from developers who show little regard for local communities.’ Shapiro faces re-election this fall. Over the past year, as opposition to such projects has mounted both within Pennsylvania and across the U.S., his stance on data centers has evolved. He repeatedly credited conversations with anxious Pennsylvania residents for his shift, stating, ‘The feedback I’ve received… has shaped my view on this issue.’ As anti-data-center sentiment intensifies amid the midterm election landscape, both Democrats and Republicans are recalibrating their positions on data center development. Opponents cite concerns ranging from rising electricity bills and environmental impacts to broader anxieties about AI technology itself. Beyond Pennsylvania, New York and Texas have also implemented explicit statewide restrictions. On July 14, New York Governor Kathy Hochul signed Executive Order No. 62, suspending state-level environmental permitting for hyperscale data centers with power capacity ≥50 MW—for up to one year. During this pause, the state will develop unified environmental impact assessment standards, protect residential electricity rates, establish frameworks ensuring community benefits, and move to eliminate sales tax incentives. A stricter bill pending in the state legislature proposes lowering the threshold to 20 MW. As a result, multiple AI supercomputing projects by Microsoft, Google, and NVIDIA have had their permitting processes paused. This marks the first statewide moratorium on AI data centers in the U.S. On August 3, Texas Governor Greg Abbott suspended approvals for new large-scale data centers and directed the Public Utility Commission and ERCOT to conduct comprehensive audits of all proposed grid-connected data centers—including assessments of electricity demand, water usage, and ownership structures. No new projects may connect to the grid until these audits conclude. Texas itself is a major AI compute hub, hosting numerous data centers developed in partnership with OpenAI—and previously pursued aggressive pro-investment policies. Now, however, it worries about grid reliability risks amid extreme weather events. Arizona, meanwhile, has adopted measures including a three-year freeze on new state-level tax incentives for data centers; mandatory utility submissions of impact reports detailing how data center interconnections affect the grid; and promotion of a ‘cost-causer pays’ mechanism—requiring large-scale compute projects to bear the full cost of grid expansion, rather than passing those costs onto residential electricity bills.

Texas Governor Abbott Halts Data Center Permits Pending Grid Audit

14 hours ago

Texas Governor Greg Abbott froze new data center permits until regulators complete an audit of energy and water consumption, impacting the state's AI-driven growth.

reuters

2026-09-21

Texas Governor Abbott Halts Data Center Permits Pending Grid Audit

Texas Governor Abbott Halts Data Center Permits Pending Grid Audit

FAQ