Singapore Is Using Access to AI Models as a Reason to Stay
The city state is betting that easy access to both US and Chinese frontier models keeps fund managers from moving to Hong Kong. Which models you can legally use now depends on where you sit.
The Financial Times reports that Singapore’s financial sector is making an unusual pitch to investment managers: stay here, and you can use whatever model you like. The argument rests on Singapore’s careful neutrality between Washington and Beijing, which means banks based there can reach both American and Chinese frontier models. Hong Kong firms increasingly cannot, because of US export scrutiny, and are countering with tax cuts instead.
The offer is not only rhetorical. Singapore has a S$150 million Enterprise Compute Initiative, essentially subsidised access to computing power for companies, and started an August pilot with around 20 financial institutions. Compute, in this context, means the specialised chips and data centre capacity needed to run or fine-tune large models, which is the expensive part of doing anything serious with AI in-house.
What is quietly significant here is the reframing. For two years the geopolitical AI story has been about chips: who can buy Nvidia hardware, who cannot, and what smuggling routes exist in between. This story is about the layer above. Which models a company is permitted to use, and under what terms, is becoming a property of the jurisdiction it sits in, not a purchasing decision. A fund manager in one city has 400 models available and one in another has 200, and neither of them chose that.
The comparison worth drawing, if you are reading this from Europe, is close to home. The EU AI Act and GDPR mean plenty of AI features launch in the United States months before they reach Germany or Austria, and some never arrive at all. That is a deliberate trade, made for reasons many Europeans support, but it is the same phenomenon Singapore is now selling against: capability that depends on your postcode. A fair caveat on the Singapore pitch itself, though. It is a marketing position as much as a policy, the pilot is small, and neutrality between two increasingly impatient superpowers is a difficult thing to maintain indefinitely.
What this means for you: if you work at a company with offices in more than one country, this is a practical thing to check rather than an abstraction. Your colleagues abroad may have access to tools you do not, and the reverse, and internal policy is often built on a stale assumption about what is allowed. If you are simply an EU-based user wondering why a feature you read about is missing from your account, this is usually the answer, and it is worth checking a provider’s regional availability before you build a workflow around something. And if you are following AI policy, watch for more of this: access to models is turning into an economic lever that governments use to attract businesses, in the same way tax rates always have.
Sources
Source: https://www.ft.com/content/7d2969fd-85aa-43fb-ae04-97d1d0ee2463
Anthropic Says It Made 11.5 Billion Dollars in Three Months, and Turned a Profit
Preliminary Q2 figures shown to investors put revenue at more than fourteen times the same quarter last year, with positive adjusted operating income. An IPO may follow in the autumn.
