Analysis
The US Banned an AI Chip, Then Unbanned It, Then Taxed It — In Twelve Months
Washington's China chip policy has swung from outright ban to a tariff-and-license framework in under a year. That volatility is arguably now a bigger risk factor than the restrictions themselves.
Within a single twelve-month stretch, US policy on a specific advanced AI chip went through four distinct positions: an export ban to China, a reversal of that ban, a 25% tariff added on top, and finally a formal licensing framework — the same product, four different regimes, inside one year.
The whiplash reflects a genuine tension inside the policy itself. A push for stable trade relations ahead of high-level diplomatic engagement pulls in one direction, while congressional pressure — including a bipartisan push for a blanket ban on semiconductor manufacturing equipment exports entirely — pulls in the other.
Underneath the policy noise, the structural trend hasn't reversed either way: China is still projected to produce only 1-2% of the AI chip volume US companies produce this year, even as it's on track to overtake Taiwan as the leading foundational chip producer by 2027 — two very different layers of the same supply chain moving in different directions at once.
Policy risk that reverses direction every few months, layered on top of a slower structural trend underneath it, is a genuinely difficult setup to hold a position through — exactly why practicing on a simulator, where a wrong read on a single headline costs nothing, is worth doing before sizing a real position around a story like this.