Alibaba opens cloud regions in Turkey and Finland as AI tensions bite
Alibaba picked the week of a US–UK AI pact to plant its cloud deeper in Europe. The same conference where it announced the regions also carried its chip roadmap — and Japan's turbine makers are still selling the power behind all of it.
Alibaba Cloud will open its first cloud regions in Turkey and Finland — plus one in the Netherlands — within the next 12 months, the company said at its Apsara conference in Hangzhou on Wednesday. The Netherlands was already announced at last year's Apsara and has not launched yet, so Turkey and Finland are the genuinely new countries; Alibaba also said it will expand existing regions in Malaysia, Germany, the UAE, France and Hong Kong. No investment figure was attached to the three new regions. Li Feifei, Alibaba Cloud Intelligence's chief technology officer and president of international business, framed it as making AI "more scalable, practical and accessible for businesses," and Bloomberg tied the push directly to escalating US–China competition in chips and models, with Trump officials pressing other governments to prefer American technology. The timing is the argument: Washington and London signed an AI pact and Alibaba announced European AI capacity inside the same news cycle, and the countries it picked are the ones with the most to gain from keeping both stacks available. Selling Chinese cloud capacity inside European jurisdictions is a harder, slower play than selling tokens — and it is the one the export controls cannot switch off.
The bigger number at Apsara was silicon: Alibaba introduced the Zhenwu V900, an AI accelerator it says is three times faster than the M890 it replaces, with mass production slated for the first quarter of 2027. CNBC reported the company is also targeting 20 GW of global data center capacity by 2032 and has Qwen 4 in training, with 5–10 trillion parameter models on the roadmap; Citigroup has estimated that scale could pull in more than $160 billion in external cloud revenue. Alibaba shares rose about 3% in Hong Kong on the day. We covered the earlier version of this plan — Alibaba plans a 10T-parameter model and a 500,000-chip cluster — and the shape has not changed: own the chip, own the cluster, then rent both out. The new regions matter less as capacity than as a distribution channel for that stack.
Mitsubishi Heavy Industries says its gas turbine order backlog is at a record high and its 2029–2030 delivery slots are already tight, with US AI data centers doing the pulling. Takao Tsukui, chairman and chief executive of Mitsubishi Heavy Industries America, told Nikkei Asia that global gas turbine demand should match or beat 2025 this year — the highest level since 2000 — and will likely clear 100 GW again, while contract prices have risen roughly 20% year on year. The company's own large-frame backlog was 35 GW in the quarter it reported in August, up from 23 GW a year earlier, and it still plans to double large gas turbine production capacity by FY2030 against FY2024, spending more than ¥100 billion (about $635 million) across Japanese and US sites. Tsukui also flagged the other half of the story: local opposition to data centers in some US states is now a midterm election issue, and delays are common. We looked at where this demand curve lands — AI data centers will out-burn Germany and Japan on gas by 2035 — and the turbine backlog is the leading indicator for all of it.
What to watch: whether Alibaba names a Turkish or Finnish launch customer. The regions are worth more as proof that Chinese AI stacks can run inside European jurisdictions than as extra capacity.
Which matters more for AI's next two years — who controls the compute, or who controls the grid that feeds it? Tell us in the comments.
