TL;DR
Alibaba CEO Eddie Wu has announced plans to train an AI model with 5 to 10 trillion parameters alongside a new proprietary AI chip, signaling China's most aggressive known push toward frontier-scale AI infrastructure.
What happened
- Alibaba CEO Eddie Wu publicly announced a target model size of 5 to 10 trillion parameters, far exceeding any publicly disclosed model today.
- The announcement paired the model ambition with an unveiling of a new in-house AI chip, suggesting Alibaba is building the full compute stack to support this scale.
- The move represents Alibaba's most ambitious AI infrastructure push to date, per the company's own framing.
- No specific training timeline or deployment date was disclosed in the announcement.
Why it matters
- Current frontier models from OpenAI, Google, and Anthropic are widely estimated in the hundreds of billions of parameters; a 5 to 10 trillion parameter model would be a 10x or greater leap over publicly known benchmarks.
- A proprietary chip alongside the model signals Alibaba is reducing dependence on Nvidia hardware, a strategic necessity given ongoing US export controls on advanced semiconductors to China.
- If achieved, this scale would position Alibaba as a credible global frontier lab, not just a regional player, reshaping competitive dynamics in enterprise AI and cloud.
- The combined model-plus-chip announcement mirrors the vertical integration playbook of Google (TPUs plus Gemini) and signals that Chinese hyperscalers are pursuing the same strategy.
- For enterprises in Asia-Pacific, this raises the prospect of a sovereign frontier model available outside US-controlled AI supply chains.
What to watch next
- Chip performance disclosures: whether Alibaba releases benchmarks for the new chip that show it can realistically support trillion-parameter training runs at scale.
- Training compute commitments: any announced data center capacity, power contracts, or capital expenditure figures that would validate the 5 to 10 trillion parameter target as near-term rather than aspirational.
- Regulatory and export control responses: whether the US tightens controls on chip-related tooling or software in reaction to Alibaba's vertical integration push.
Originally published on Present of AI, a daily source-linked AI news timeline. Read the full timeline or browse the open dataset.