Alibaba has unveiled the Zhenwu V900, a new AI accelerator that pushes the company deeper into the hardware behind artificial intelligence. Announced at the 2026 Apsara Conference in Hangzhou, the processor arrives alongside a much larger AI infrastructure plan spanning chips, cloud data centers and future Qwen models.

Alibaba’s Zhenwu V900 targets AI training and inference

According to Alibaba Cloud’s official announcement, the Zhenwu V900 was developed by T-Head for both AI training and inference. Alibaba says the accelerator delivers three times the performance of the earlier Zhenwu M890 and includes 216 GB of memory, 1,200 GB/s of inter-chip bandwidth, plus native FP8 and FP4 support.

The hardware is scheduled for mass production and commercial release in the first quarter of 2027. Independent reporting from Reuters and The Associated Press also covered the launch this week, making the announcement one of the day’s major AI hardware stories.

Why the 500,000-chip cluster matters

Alibaba is not positioning the V900 as a standalone processor. Its upgraded supernode server combines the accelerator with the company’s ICN Switch, Panmai SmartNIC and Zhenyue SSD controller. Alibaba says the architecture can scale to clusters containing as many as 500,000 accelerator cards.

That system-level approach matters because modern AI infrastructure is increasingly constrained by more than raw compute. Memory capacity, chip-to-chip bandwidth, storage and networking all influence how efficiently very large models can be trained and served. The V900 announcement therefore represents a broader AI hardware story rather than simply another chip launch.

Alibaba plans more than 20 GW of data-center capacity

The chip reveal came with an unusually large infrastructure target. Alibaba CEO Eddie Wu said the company wants Alibaba Cloud’s global data-center capacity to exceed 20 GW by 2032. That makes power delivery, networking, cooling and server density increasingly important parts of the AI race—areas closely connected to the data-center hardware coverage followed by BitcoinVersus.Tech.

Alibaba also said Qwen 4 is in training and outlined future Qwen 4.5 and Qwen 5 models that could scale to roughly 5 trillion to 10 trillion parameters. Parameter count alone does not determine model quality, but the roadmap helps explain why Alibaba is simultaneously investing in accelerators, interconnects, storage and enormous data-center capacity.

Watch: Apsara Conference 2026

The official Alibaba Cloud video above introduces Apsara Conference 2026, the event where the Zhenwu V900 and the broader full-stack AI roadmap were announced.

What to watch next

The most important next milestone is real-world deployment. Alibaba’s three-times-performance figure is a company claim relative to its own predecessor; independent workload benchmarks will provide a clearer picture once V900 systems become commercially available. The announced Q1 2027 production window, customer deployments and power efficiency at scale will be key measures to watch.

---

BitcoinVersus.Tech Editor’s Note:
We volunteer daily to ensure the credibility of the information on this platform is Verifiably True. If you would like to support to help further secure the integrity of our research initiatives, please donate here: 3C9o19EH5HSiwEPyCTmEKzxhNCbo2X6TTb