Alibaba Targets 20GW of Data Center Capacity by 2032 and Unveils Zhenwu V900, "China's Most Powerful AI Chip"
At its annual Apsara Conference (announced ~September 22, 2026), Alibaba CEO Eddie Wu said the company will expand global data center capacity to more than 20 gigawatts by 2032 to meet AI demand, and unveiled the Zhenwu V900 AI chip — which Alibaba described as "the most powerful AI chip in China," capable of "scaling up to 500,000 cards per cluster." Alibaba also plans Qwen models with 5–10 trillion parameters "as part of its efforts towards artificial superintelligence," and Wu framed the opportunity by saying machine thinking "currently represents less than three per cent of human capacity." The company is investing across three core areas: AI models, chips, and cloud infrastructure, building toward an "AI cloud designed for the Machine Intelligence era."
Why it matters. This is the supply-side pressure behind this week's US price war: CNBC explicitly named Alibaba (alongside Moonshot AI and DeepSeek) as the open-weight competition pushing Anthropic and OpenAI to slash prices (OpenAI Answers Opus 5.5 Same-Day with GPT-6 Sol and Luna at 50% Cuts — the Price War Moves to Cost-per-Task, Xiaomi's MiMo-V2.6-Pro Takes the Open-Weights Crown — Tying Grok 4.7 on a Trillion-Parameter MoE Trained on Chinese Chips). It also escalates China's domestic-silicon push one week after Huawei's Ascend 960 announcements (Huawei Doubles Down on Domestic AI Silicon: Two Ascend 960 Chips Slated for 2027 Despite US Curbs) — and a 20GW-by-2032 target is a hyperscaler-scale commitment that competes directly with the US labs' gigawatt data-center land grabs. Caveats: the 500K-card cluster and "most powerful in China" claims are vendor-stated, and the detailed figures here rest on conference reporting (ET Datacenters) rather than a primary Alibaba document.