By Brian Newman, Contributing Editor
Insight at the Intersection
The November 3 midterms moved control of both chambers of Congress to Democrats, and the AI infrastructure trade repriced before the final counts were in. Reuters reported that cash-rich AI heavyweights rose the next day while chipmakers and data center operators fell, and the Philadelphia Semiconductor index touched a six-month low. The sharper signal came from the states. In Texas, governor-elect Gina Hinojosa campaigned on a statewide moratorium on data center approvals until the Legislature passes new protections.
The exposure is concentrated. Modo Energy found that 165 GW of the 633 GW of data center capacity it tracks, about 26 percent, sits in states whose incoming governor backs an immediate moratorium.
Exit polls showed a majority of voters calling AI data centers a crucial or major factor in their vote. The pressure is not confined to one party: Republican Governor Greg Abbott paused Texas data center permits earlier this year, and the House passed the Ratepayer Protection Act 417-3 in September.
The United States still owns the chips. The election raised the price of power and permits, the one front where China was already ahead.
The next question is whether this hands China an advantage, and the honest answer is narrower than the headlines suggest. On compute, the American lead is wide. Epoch AI’s customs-based estimates show the United States received over 70 percent of new AI compute delivered since 2022, against about 10 percent for China. Epoch also estimates that Huawei will produce less than 4 percent as much AI compute as Nvidia in 2026.
China’s position looks different on two other fronts. Inside its own market, Huawei reached 28 percent of China’s data center AI chip market in the first half of 2026 against Nvidia’s 34 percent, according to IDC data reviewed by Reuters.
On power, China already generates more than twice as much electricity as the United States. BloombergNEF projects that data centers will drive 38 percent of U.S. electricity demand growth between 2024 and 2030, compared with 6 percent in China.
I believe the election did not hand China the lead in AI, although it did hand China time on the one dimension where it competes best. The reasoning runs in four steps. First, the U.S. advantage rests on chips and capital, and neither changed on November 3. Second, the binding U.S. constraint has shifted from silicon to siting, and the election made siting slower and more expensive in the states carrying the largest share of the buildout. Third, China’s binding constraint is memory and advanced manufacturing, and the incoming Congress is more likely to tighten that constraint than relax it. Finally, a race between a chip-constrained challenger and a permit-constrained leader becomes a contest over which bottleneck clears first. As a result, any advantage China gains is relative and conditional rather than structural.
Two Bottlenecks, One Race
Export controls are where the incoming Congress is most likely to act, and on this issue the hawks sit in both parties. Representative Jim Himes, the top Democrat on the House Intelligence Committee, has called chip restrictions the key lever for the U.S. edge over China. The House passed the Remote Access Security Act 369-22 in January to extend export controls to cloud rental of advanced chips, and the bill still awaits Senate action. Analysts expect a Democratic-led Congress to increase pressure on Beijing over AI and export controls.
Those controls bite hardest on memory. Chinese AI chipmakers raised prices this fall as high-bandwidth memory supply tightened, with TrendForce reporting increases of up to 50 percent on Huawei’s 950DT. China’s leading DRAM maker, CXMT, is sampling HBM to domestic chip developers and targeting HBM3 mass production by the end of 2026. That milestone deserves close attention, because it would loosen China’s constraint at the same moment the U.S. siting constraint is tightening.
Telecom sits in the middle of both stories. SpaceX’s plan to become a hybrid satellite-terrestrial carrier depends on FCC approval of its 800 MHz purchase from Grain Management, and this week Qualcomm agreed to supply equipment for tens of thousands of Starlink Mobile cell sites starting in 2027. A Congress holding committee gavels and subpoena power is likely to look harder at spectrum transfers, which raises execution risk without stopping the technical momentum. Incumbent carriers should expect the same closer oversight of spectrum and federal broadband programs that their new competitor faces.
Years in network reliability reinforce a basic engineering rule: capacity concentrated in a few sites is capacity at risk, and diversity of route and location is how networks survive a single point of failure. The AI buildout concentrated a large share of its planned capacity in a handful of states, and the election exposed that concentration. Carriers own thousands of central offices, switching centers, and edge sites that are already powered, permitted, and connected to fiber. If hyperscale campuses stall, inference workloads gain a reason to move toward distributed capacity, and telecom operators hold much of the real estate that makes distribution possible.
The most likely outcome is a slower and costlier U.S. buildout through 2027, with tighter chip controls preserving the compute lead. Some of the alarm may prove overstated, since SemiAnalysis estimated that only about 2.3 GW of U.S. capacity was genuinely delayed by local moratoriums before the vote, against roughly 38 GW of new capacity projected for 2027. On the contrary, if statewide moratoria spread beyond Texas and CXMT reaches volume HBM production, China’s window widens faster than export controls can close it.
Will the United States clear its permitting bottleneck before China clears its memory bottleneck? For telecom operators sitting on powered and permitted sites, is the slowdown a threat to their largest customers or the opening the industry has waited a decade to claim?
Brian C. Newman is a telecom and AI strategy consultant, course creator, and former Verizon technology leader with more than 30 years of experience across wireless networks, 5G, network operations, infrastructure modernization, and emerging technologies. He helps organizations understand how AI, connectivity, edge computing, and digital infrastructure are reshaping business operations, real estate, public safety, and customer experience.
