For three years the AI race has been scored on capability — bigger models, higher benchmarks, faster inference. This week the more telling movement was on a different axis entirely: not what the systems can do, but where they are permitted to do it. Two stories, pointing in opposite directions across the same US–China divide, made the case that market access is becoming as decisive as model quality.

On Friday, Apple was reported to have cleared a regulatory bar no foreign company had cleared before. The same week, a Chinese self-driving firm booked itself a route into Europe. The border between the two largest AI markets is not closing so much as being crossed — carefully, and on terms set by regulators rather than engineers.

Apple wins a China first

Apple has trained its own large language model for the Chinese market, with help from Alibaba, rather than simply renting a domestic model to power Apple Intelligence there. According to Reuters, that makes Apple the first foreign company approved by the Chinese government to offer a proprietary AI model in the country.

The clearance is concrete, not rumoured. China's Cyberspace Administration registered Apple's generative-AI service in July, listing Apple Technology Development (Shanghai) as the entity — by Reuters' account, the first such clearance for a foreign firm's own model. The self-trained system is expected to sit alongside Alibaba's Qwen, which is also being woven into the China edition of Apple Intelligence; the features are expected to reach Chinese iPhones in the coming months after an iOS update.

The significance is less about the model than the precedent. Beijing has spent years steering foreign firms toward domestic models and keeping proprietary Western AI at arm's length. Apple has its own reasons to want a credible offering: its iPhone position in China has been under well-documented pressure from domestic rivals shipping AI features of their own. Letting Apple run its own model — even one trained with a local champion's help and shipped beside Qwen — is a notable concession, and one every other Western company eyeing the Chinese market will now cite.

A Chinese fleet drives into Europe

Pointing the other way: Pony.ai and Uber have expanded their partnership to put more than 2,000 of Pony.ai's robotaxis onto Uber's platform across Europe, rolling out in phases. The existing service in Zagreb moves onto Uber, with four further European cities to follow and deployments in the Middle East also planned. Pony.ai supplies the Level 4 self-driving technology and operations; Uber provides the app, payments and riders; local partners handle the day-to-day fleets.

Europe has lagged the United States and China on robotaxis, so a scaled commitment there is notable on its own. That it is a Chinese autonomy firm doing the scaling — into Western cities, on a Western platform — is the mirror image of the Apple story: Chinese AI moving into the West at the same moment Western AI is being let into China.

Meanwhile, the money didn't pause

If there was any doubt the capital behind all this remains undimmed, Databricks raised a fresh $5 billion this week at a $190 billion valuation — only weeks after it was marked at $188 billion in mid-July. The company says it is now at a $7 billion revenue run-rate, up more than 80% year on year, with the money aimed at its agent-database, data-assistant and AI-governance products. The frontier labs get the headlines; the infrastructure underneath them is still raising in five-billion-dollar increments.

Taken together, the week's news redraws the map a little. The question is shifting from whose model is best to whose model is allowed where — and the answers are being written in regulatory registries and cross-border deals as much as in benchmark tables.