The BBC reports that new US rules on South East Asia restrict chip exports and also access to the computing power those chips provide, which China relies on heavily. This guide shows you how to check whether your own compute supply chain is affected.
What you need before starting
You need an up-to-date list of every computing service your organisation uses or sells. That includes cloud accounts, colocation contracts, managed AI platforms and any hardware you own in data centres abroad. You also need copies of the contracts behind those services and a list of the customers or partners who use your capacity. Ideally, someone in the team should already know export compliance.
Know the limits of what has been published so far. The BBC’s report says the rules cover chip exports and access to computing power, and that China relies heavily on that access. The material available here does not say which South East Asian countries are covered, which chips fall within scope, what thresholds apply or when the measures take effect. Do not fill those gaps with guesses. Get the official text from the US authorities before you draw firm conclusions.
Understand the two channels being closed
US export controls on advanced semiconductors have usually focused on physical goods: a chip or server leaving one country for another. That approach left two openings. Hardware could be shipped to a third country and then moved on. A company could also skip buying hardware altogether and rent processing time on servers in another country, reaching them over the internet.
According to the BBC, the new rules deal with both openings in South East Asia. The second matters most for China. If Chinese developers cannot buy high-end chips directly, renting capacity in nearby data centres is an obvious alternative, and the BBC says China leans heavily on that kind of access. Limiting it targets compute as a service, not just the silicon. How much this will slow Chinese AI work is not known. It will depend on how the rules are enforced, how quickly alternatives appear and how capable domestic Chinese chips become.
Map where your computing power physically runs
Start with geography. For each service on your list, record the country where the servers actually sit, not where the provider has its headquarters. Many large cloud providers let customers choose a region, and a single account can run workloads in several countries at once. Check the deployment settings, not just the sales contract.
Mark every workload that runs in South East Asia. If you are unsure where a managed service processes data, ask the provider in writing and keep the reply.
Identify the hardware behind each service
Export controls on semiconductors usually depend on the performance of the chip. They do not cover all computing. A general-purpose web server and a cluster of high-end accelerators built for training AI models are likely to be treated very differently. For each workload you flagged in South East Asia, find out what processors it runs on.
Providers often list this in instance or product descriptions. If you own the hardware, check the asset register and purchase records. Until the official thresholds have been confirmed, record the exact model so you can compare it against them later.
Check who ultimately uses the capacity
The BBC’s report puts the focus on Chinese access to compute, so end users matter as much as hardware. If you only consume compute for your own staff, this step is short. If you resell, sublet or give partners access to capacity, list who those parties are, where they are based and who owns them.
Pay attention to indirect relationships. A customer registered in one country may be owned by, or working for, an entity in another. Your existing know-your-customer and sanctions-screening processes are the natural place to record this. Note any accounts where ownership is unclear.
Review contracts and access controls
Read your contracts with providers and customers for clauses on export control, sanctions and acceptable use. Many cloud agreements already require customers to follow export laws and allow the provider to suspend service. Find out who carries the obligation in each relationship and what notice periods apply if access has to be cut.
Then check the technical side. Identify who can create accounts, start high-performance instances or move workloads between regions. Logging that records who used which resources, and from where, will help if you later need to show compliance or investigate misuse. Tightening identity and access management is good security practice anyway, and it now has a regulatory reason too.
Monitor official guidance and record your decisions
Export rules are often clarified after publication through guidance, licence decisions and frequently asked questions from the issuing authority. In the United States, the Commerce Department’s Bureau of Industry and Security administers the Export Administration Regulations. Assign someone to watch for updates and to compare each one against your inventory.
Write down what you checked, what you found and what you decided, with dates. If a regulator, customer or auditor asks later, a clear record of reasonable steps is far more useful than a reconstruction from memory.
Mistakes people actually make
The most common mistake is assuming that cloud services fall outside export controls because no physical item crosses a border. The BBC’s description of the new rules makes clear that access to computing power is part of the target, so that assumption should be dropped.
A second mistake is judging exposure by the provider’s nationality rather than the data centre’s location. A familiar provider can still run your workload in a covered country.
Third, organisations often accept a reseller’s general statement of compliance without checking the hardware or the end users themselves. Responsibility does not always pass down the chain.
Fourth, people mix up earlier US semiconductor restrictions with the new measures. Rules have been revised several times, and an assessment written for a previous version may not reflect the current one.
Finally, teams often check only the parent company and forget subsidiaries, joint ventures and research partners, which may hold their own cloud accounts.
When this approach is the wrong choice
This checklist is for organisations that use, sell or host computing capacity and want a first view of their exposure. It does not replace legal advice. If you export hardware directly, operate data centres in the region or serve customers with links to China, you need specialist export-control counsel and probably a formal compliance programme.
It is also the wrong tool if you want to know what the rules mean for geopolitics or the global chip market. Those questions call for policy and economic analysis, not an inventory exercise. And if your organisation is based in China, other considerations apply, including Chinese law and possible countermeasures, which this guide does not cover.
Frequently asked questions
What do the new US chip rules on South East Asia cover?
According to the BBC, the new US rules restrict exports of chips to South East Asia and also restrict access to the computing power those chips provide. The publicly available summary does not set out which countries, chip models or performance thresholds are included, or when the measures apply. Anyone who needs those details should consult the official regulatory text published by the US authorities and any guidance issued alongside it.
Why does South East Asia matter for China’s access to AI chips?
South East Asia hosts a growing number of data centres and is geographically close to China. When direct purchases of advanced chips are restricted, the region offers two possible workarounds: moving hardware on through third countries, or renting computing capacity remotely. The BBC reports that China relies heavily on access to computing power, which explains why US policy is now looking at the region rather than only at direct sales to China.
Does renting cloud computing count as an export under US chip rules?
Traditionally, export controls have focused on physical goods and technology transfers, which left remote access to computing power in a grey area. The BBC reports that the new rules covering South East Asia restrict access to computing power as well as chip exports. Exactly how cloud rental is defined and treated in the new measures is not set out in the available summary, so organisations should check the official text and seek legal advice.
Will the US restrictions stop China developing advanced AI?
It is not known how much the restrictions will slow China’s AI development. The effect depends on how strictly the rules are enforced, whether other regions offer alternative computing capacity, and how quickly Chinese domestic chip production improves. The restrictions make it harder to get high-end computing power through South East Asia, which the BBC says China relies on heavily, but closing one channel does not necessarily close every route.
Sources and further reading
- BBC News technology coverage, which reported that the new rules restrict both chip exports and access to computing power
- US Department of Commerce, Bureau of Industry and Security, which publishes export control rules and related guidance
- The Export Administration Regulations, the US legal framework under which semiconductor export controls are issued
- Academic and think-tank analysis of semiconductor export controls and the cloud computing gap
Surfaced from the rss:bbc_tech signal “US chip export restrictions”. AI-assisted draft, editorially reviewed.

