Bipartisan former US officials urged a grid overhaul to win the AI race with China.
Former senior officials who served in past U.S. administrations have crossed party lines to urge an expansion of energy infrastructure. They argued that to avoid falling behind China in advanced industries such as artificial intelligence (AI) and semiconductors, the United States must streamline complex permitting processes and speed up the build-out of power grids and generation facilities.
On Oct. 20 (local time), The Wall Street Journal (WSJ) reported that former senior officials who served in the administrations of Donald Trump, Joe Biden, and Barack Obama urged reforms to permitting procedures that block energy infrastructure construction. They did so in a joint letter sent that day to U.S. congressional leadership.
Signatories included former Secretary of State Mike Pompeo from Trump's first-term administration; former Treasury Secretary Janet Yellen and former National Security Adviser Jake Sullivan from the Biden administration; and former Council of Economic Advisers Chair Jason Furman from the Obama administration. Steven Mnuchin, a former Federal Reserve (Fed) governor appointed by President Trump, also joined.
The signatories noted that while the United States has strengths in technology, capital markets, talent, and natural resources, it will be difficult to maintain this edge if energy infrastructure build-out is delayed. If power cannot be supplied in time to AI data centers, semiconductor plants, and facilities related to critical minerals, investment and production in advanced industries could be disrupted.
They also raised the issue of differing speeds in transmission grid expansion between China and the United States. According to the letter, the scale of high-voltage transmission lines China has built over the past 15 years exceeds the United States' total to date. By contrast, the United States faces difficulties constructing large-scale energy facilities due to complex environmental reviews, permitting, and years-long litigation.
The letter urged swift passage of the Bipartisan American Affordability and Jobs Act of 2026, which Democrats and Republicans jointly introduced in the U.S. Senate last month. Released on Sept. 30, the bill would streamline environmental reviews and permitting for energy facilities and promote expansion of the transmission grid. It also includes a provision requiring AI data centers to bear the expense of building new transmission lines and related costs.
These calls come as transmission expansion fails to keep pace with the surge in AI data center construction in the United States. According to the International Energy Agency (IEA), data centers can typically be built in two to three years, but in advanced economies it can take four to eight years to build new transmission lines. The IEA analyzed that if grid bottlenecks are not resolved, about 20% of planned data center projects worldwide risk delays.
Delays and cancellations of data center construction are already increasing in the United States. The Financial Times (FT) reported on Oct. 18, citing data from regulatory intelligence firm RegRayr, that at least 47 large data center projects in the United States have been delayed or withdrawn this year, and some have been blocked from approval. Investment bank Bernstein estimated that only about 35% of planned U.S. data center projects are likely to materialize.
However, there is considerable opposition to easing permitting regulations. While the oil and gas and renewable energy industries support the bill in hopes of boosting energy infrastructure investment, 185 environmental, labor, and community groups sent a letter of opposition to Congress on Oct. 13. They argued the bill could weaken environmental safeguards and trigger unchecked expansion of large AI data centers and fossil fuel-based power plants.
Former officials worry that the longer such debates delay the bill, the more U.S. competitiveness in advanced industries could erode. Delays in building out the grid could push back the start dates for operating data centers, affecting delivery schedules for the semiconductors and memory that go into AI servers.