The strategy was clean on paper. America would win the global AI race by controlling access to the chips that power it. Through export licences and licensing requirements, the US government could determine who got to build serious AI infrastructure and who didn't, keeping the most capable chips away from strategic adversaries while monetising access for allies and partners. It was, in theory, a masterstroke of geopolitical leverage: turning America's semiconductor dominance into a foreign policy instrument.
The theory is meeting a different kind of obstacle. Not China, not allied resistance, not a legal challenge. A staffing shortfall at the Commerce Department's Bureau of Industry and Security, the agency responsible for processing chip export licences, has created a bottleneck that is now tangibly impeding the very chip sales the strategy was designed to facilitate. A Bloomberg investigation published in mid-April found the bureau drowning in applications it cannot process, dealing with staff attrition at precisely the moment when application volumes are surging, and receiving inconsistent policy direction from above. Major rules have been proposed, withdrawn, revised, and proposed again within weeks, leaving companies trying to obtain licences with no reliable timeline for approvals.
Some firms have given up waiting entirely and are adjusting their supply chains to route around the bottleneck, not always in directions that favour American manufacturers. Nvidia and AMD, which are supposed to be the primary beneficiaries of America's AI hardware advantage, find themselves watching sales stall in markets where buyers have lost patience with the permit process and started exploring alternatives.
Congress has added its own layer of complication. A Washington Post tech brief in early April reported legislators pushing for additional crackdowns on chip manufacturing equipment exports, creating a separate stream of policy pressure on the same overloaded regulatory apparatus. The result is an agency trying to implement a constantly shifting mandate with a shrinking workforce, under increasing political scrutiny from multiple directions simultaneously.
There is a historical pattern worth noting here. During the encryption export wars of the 1990s, the US government classified strong encryption as a munition and required export licences for software that used it. The rationale was national security: keeping strong encryption out of adversaries' hands. The practical outcomes were: adversaries developed their own strong encryption anyway, because the underlying mathematics is public knowledge; American software companies lost international market share to European competitors who faced no such restrictions; and the technology proliferated regardless, on a timeline that the export controls did not meaningfully alter. The controls were eventually unwound after roughly a decade, by which point the damage to American software competitiveness was not recoverable.
Chip export controls are not a perfect analogy. Physical manufacturing capacity is harder to replicate than a mathematical algorithm, and genuine lead times in semiconductor manufacturing can matter strategically. There are probably specific cases where delaying a particular adversary's access to a particular chip generation by even a few months changes a real calculation. The argument for some form of export control is not absurd.
But the argument for the current situation, where the bureaucracy processing applications is overwhelmed, the policy direction keeps shifting, and friendly-nation companies are being driven to seek alternatives, is much harder to make. The administration is capturing none of the strategic upside and absorbing all of the administrative cost. A foreign policy instrument that primarily slows down your own exporters is not leverage; it is self-inflicted friction.
The fix, in principle, is straightforward: resource the agency, stabilise the policy, and give companies predictable timelines. Whether the political will to do that exists, in an administration where the export control regime is also serving as a domestic political signal, is a different question.