Table of Contents
Alibaba CEO Eddie Wu said the company aims for the global data center capacity operated by Alibaba Cloud to surpass 20GW by 2032
In sum – what to know
Full-stack AI push – Alibaba is expanding its strategy across models, proprietary chips, cloud infrastructure and agent platforms as it prepares for rising AI demand.
20GW capacity target – Alibaba Cloud aims to surpass 20GW of global data center capacity by 2032, a target Omdia describes as ambitious and challenging.
Chips hedge GPU constraints – Alibaba’s proprietary AI chips are becoming an important part of its infrastructure strategy, including as a hedge against increasingly scarce access to Nvidia GPUs.
Chinese company Alibaba is expanding its full-stack AI strategy across models, proprietary chips, cloud infrastructure, and agent platforms, while setting an ambitious target to expand the scale of its global data center operations by 2032.
At the Apsara Conference in Hangzhou, Alibaba said its next-generation Qwen 4 model is currently in training and outlined plans for the Qwen 4.5 and Qwen 5 series, which are projected to scale to between 5 trillion and 10 trillion parameters.
The company also unveiled the Zhenwu V900, a new AI training and inference processor developed by its T-Head chip design unit. Alibaba said the V900 delivers three times the performance of the Zhenwu M890, with 216 GB of GPU memory and 1,200 GB/s of inter-chip bandwidth. The chip is scheduled for mass production and commercial release in the first quarter of 2027.
T-Head’s Zhenwu AI chips are already serving more than 650 customers, according to Alibaba. The company also unveiled an upgraded supernode server combining the Zhenwu V900 with its ICN Switch, Panmai SmartNIC and Zhenyue SSD controller chip, with support for supernode clusters comprising up to 500,000 cards.
The hardware expansion is part of a broader infrastructure strategy. Alibaba Cloud unveiled upgrades spanning AI model training and inference, networking, storage, agent deployment, and data management.
Its AI Native Cloud includes an upgraded Platform for AI and a new Cloud Parallel File Storage system designed for AI workloads. Alibaba said the storage system delivers hundred-terabyte-per-second throughput and hundred-million IOPS, while reducing enterprise AI storage costs by 69%.
Alibaba also introduced HPN 8.0 Pro, its latest AI networking architecture, which provides 100 petabits of bandwidth and supports more than 130,000 network ports at 800G speeds in a single cluster.
The Chinese company is also developing an Agent Native Cloud centered on AgentCore, a platform for building, running, and managing AI agents, alongside an Agent Security Center for security and compliance. Its Context Engine provides AI agents with real-time context and long-term memory, while Alibaba said its Agent Context service can reduce token usage by up to 67%.
The infrastructure roadmap is backed by a major data center expansion target. Alibaba CEO Eddie Wu said the company aims for the global data center capacity operated by Alibaba Cloud to surpass 20GW by 2032.
Omdia chief analyst Lian Jye told RCRTech that reaching that level would require Alibaba to add about 2–3GW of capacity annually over six years, which would be faster than its historical pace.
“Reaching >20 GW in six years implies adding about 2–3 GW per year on average—faster than Alibaba has historically added. It is very ambitious and therefore challenging,” the analyst said.
Lian Jye also pointed to potential constraints on the expansion, including chip supply and shortages of advanced AI data center infrastructure components, as well as the significant capital requirements associated with sustained infrastructure investment.
“This suggests Alibaba Cloud has strong confidence in future market demand for AI infrastructure, a sentiment commonly shared by key industry vendors,” he said.
The analyst said the broader roadmap continues Alibaba Cloud’s full-stack AI strategy, with the company seeking to strengthen its portfolio of models, silicon and infrastructure.
“The roadmap is a continuation of Alibaba Cloud’s full-stack AI strategy. As the company aims to become the AI vendor with the most powerful full-stack AI solutions, it needs to keep strengthening its portfolio of models, silicon, and infrastructure,” he said.
Lian Jye also highlighted the role of proprietary chips in Alibaba’s infrastructure strategy. He said AI cost measurement is increasingly shifting toward metrics including time to first and last token and inference speed, making an optimized infrastructure stack increasingly important.
