Lines of Thought / Topic
Export controls: the line so far
13 developments and 4 rules, in date order. Built automatically from everything tagged with this topic.
- Rule · In forceAENational AI Strategy 2031
- DeepSeek releases R1 reasoning model with open weights under MIT licence
It showed that near-frontier reasoning could be reproduced cheaply and given away openly, resetting price expectations and sharpening US debate on export controls and open weights.
- US scraps AI Diffusion Rule and warns against using Huawei Ascend chips
It replaced a single global framework for AI-chip exports with country-by-country deals, which is why Gulf and China access since then has been negotiated case by case.
- Stargate UAE: 1GW OpenAI cluster planned in Abu Dhabi, first 200MW due in 2026
It tied US chip access to sovereign AI investment deals, a template for later US approvals of advanced-chip exports to the Gulf.
- China's internet regulator summons NVIDIA over alleged security risks in H20 chips
It showed Beijing would use security reviews to slow purchases of US chips that Washington permits, pushing Chinese buyers towards domestic accelerators.
- China's market regulator finds preliminary NVIDIA antitrust breach over Mellanox deal
It gives Beijing a pending penalty lever over the leading AI-chip supplier that can be tightened or eased with the wider trade relationship.
- Huawei sets yearly Ascend AI-chip roadmap and 8,192-chip Atlas 950 SuperPoD
It is China's main domestic answer to US export controls, betting on very large clusters of weaker chips to close the per-chip gap.
- China adds extraterritorial rare-earth export controls, then pauses them to Nov 2026
Rare earths feed chipmaking tools, electronics and power equipment, so the November 2026 expiry is a live supply risk for data-centre and chip supply chains.
- US opens case-by-case H200 licences for China and adds 25% tariff on such chips
It reopened China to near-frontier US accelerators under a revenue-and-volume bargain, reversing the trend of tighter controls since 2022.
- NVIDIA restarts H200 production after Chinese orders and first US licence
Both governments now gate the same chip sales, so China's AI compute supply depends on two sets of approvals that can be withdrawn at any time.
- US charges Supermicro co-founder over alleged $2.5bn Nvidia-server diversion to China
It is the largest alleged AI-chip diversion case yet and puts server makers' end-use checks, not just chip designers', under enforcement scrutiny.
- US export-control order forces global shutdown of Claude Fable 5 and Mythos 5
The US used export controls to halt a commercial frontier model, showing that cyber-capable models can be pulled from every customer overnight.
Also on When models learned to hack
- US moves UAE into top export tier, opening licence-free AI chips to approved users
It codifies the approved-entity model for exporting the US AI stack to partners, and lawyers expect it to be the template for any other Gulf or strategic partner given similar treatment.
- OpenAI ties large reasoning-distillation campaign to people linked to Moonshot AI
It puts distillation by rival labs on the record as a security threat, likely feeding US debate on restricting Chinese access to American model APIs.