Export controls hitting bottlenecks
U.S. efforts to expand AI chip exports are being slowed by staffing and licensing problems inside the Commerce Department, creating delays in the licensing process. Reporting says the export‑control bureau is strained by attrition and capacity issues while enforcement questions persist after a Chinese firm disclosed roughly $92M of restricted Nvidia servers to Beijing ( ). The result is more policy uncertainty for supply planning: rules are tightening on paper even as administrative and enforcement gaps make outcomes unpredictable (bloomberg.com).
The United States is writing tougher rules for artificial intelligence chips at the same time its own licensing office is falling behind, so companies can see the stop signs but still can’t tell which trucks will actually be waved through. Transport Topics, citing Bloomberg reporting on April 10, said the Commerce Department’s export bureau has staffing losses, longer reviews, and a growing backlog for chip licenses. (ttnews.com) The office at the center of this is the Bureau of Industry and Security, which is the part of the Commerce Department that decides who can ship sensitive technology abroad. Transport Topics said staff turnover is near 20% and processed licenses are down about 25%, which turns a policy tool into a paperwork bottleneck. (ttnews.com) An export license is basically a government permission slip for a sale that would otherwise be blocked or tightly controlled. When reviews stretch from weeks into months, chipmakers, cloud companies, and server makers can’t plan factory output or delivery schedules with much confidence. (ttnews.com) These controls matter because the chips in question are not ordinary computer parts. The January 15, 2025 “Framework for Artificial Intelligence Diffusion” rule from the Bureau of Industry and Security targeted advanced computing chips and even some artificial intelligence model weights, treating them as strategic assets with military and surveillance risk. (federalregister.gov) Then the policy itself shifted. On May 12, 2025, the Commerce Department said it was rescinding the Biden-era artificial intelligence diffusion rule while adding other semiconductor export steps, which meant companies had to adjust to a new map even before they knew how fast the permit office could process the route. (bis.gov) That would already be messy, but enforcement questions got sharper this week. Bloomberg reported on April 8 that a Shenzhen company called Sharetronic disclosed roughly $92 million of server purchases containing restricted Nvidia H100 or H200 processors in filings to Beijing, despite United States limits on those systems. (bloomberg.com) Bloomberg also reported that the disclosure came hours after the United States charged a Super Micro Computer co-founder with illegally smuggling billions of dollars’ worth of Nvidia artificial intelligence chips to China. That pairing made the problem look less like a single paperwork lapse and more like a system where formal rules and real-world movement are drifting apart. (bloomberg.com) Nvidia has spent years redesigning products to fit inside export limits, while Washington has kept tightening the thresholds those products must meet. Transport Topics said the current delays are now colliding with a White House push to expand approved exports, which leaves companies stuck between commercial pressure to sell and national security pressure to slow down. (ttnews.com)
