Alibaba CEO: Scaling Top-Tier AI Thinking to Redefine Rare Disease R&D

At the 2026 Apsara Conference, Alibaba Group CEO Wu Yongming outlined the company’s strategic vision for the emerging machine intelligence era, declaring that artificial intelligence is transitioning from a productivity tool to a primary driver of computational and scientific discovery. Wu emphasized that current generative applications represent only the earliest phase of a broader technological shift, predicting that machine-generated reasoning could eventually exceed human cognitive output by a factor of 1,000. He argued that as computational costs decline and model capabilities mature, elite intellectual resources will become scalable commodities, fundamentally reshaping capital-intensive domains such as rare disease research and complex multi-agent engineering. To sustain this transition, Alibaba committed to a tripartite infrastructure strategy centered on foundational AI models, specialized silicon, and next-generation cloud computing. The Qwen research division will advance toward recursive self-improvement and integrated multimodal capabilities, with active plans to train next-generation models scaling between five trillion and ten trillion parameters. On the hardware front, Alibaba’s T-Head division unveiled the Zhenwu V900 AI accelerator, delivering three times the performance of its predecessor and enabling single-cluster configurations of up to 500,000 chips for large-scale model training. The company also confirmed that its existing M890 super node architecture is already supporting inference workloads for models exceeding two trillion parameters, with commercial cloud availability expanding this quarter. Complementing silicon advancements, Alibaba projected a massive expansion of its AI cloud footprint, targeting more than 20 gigawatts of global data center capacity by 2032. This infrastructure shift aims to decouple intelligent computing from traditional internet workloads, providing developers and enterprises with on-demand access to persistent AI agents and continuous reasoning pipelines. Wu drew parallels to the early electrical grid, noting that while initial applications merely replicate existing human workflows, the true economic breakthroughs will emerge from entirely new product categories yet to be invented. He concluded that by offloading repetitive cognitive labor to machines, society can reallocate human attention toward creative exploration, marking the beginning of a new industrial paradigm and signaling Alibaba’s decisive pivot toward foundational AI infrastructure.