
Alibaba used its September 22 developer conference to introduce the Zhenwu V900 accelerator, saying the new chip delivers roughly three times the performance of its predecessor. The company said mass production is scheduled for early 2027.
Alibaba Group Holding Limited (NYSE:BABA) also outlined a broader AI push. That includes future Qwen models that, according to the company, could scale to 5 trillion to 10 trillion parameters, as well as a plan to expand global data-center capacity to more than 20 gigawatts by 2032.
Alibaba is building a vertically integrated AI stack
The article argues that Alibaba’s advantage lies in being able to deploy a homegrown accelerator across its cloud platform, models and internal applications at the same time. That kind of vertical integration gives the company built-in demand before the chip is sold widely in the market.
If Zhenwu performance improves further and demand for Qwen rises, Alibaba could capture more cloud revenue without relying on an outside GPU supplier for every added unit of compute. The 20-gigawatt capacity target also suggests the company is preparing for a much larger infrastructure footprint.
At the same time, the story highlights clear risks. The V900 performance claims come from Alibaba itself, while mass production is not expected until 2027. Strong benchmark results do not necessarily eliminate risks tied to manufacturing yields, software tooling, memory supply or adoption by developers.
Alibaba is also investing aggressively in AI while its core commerce business is still expected to generate the cash needed to fund that buildout. In other words, the expansion depends on the cash flow of the company’s existing operations.
Nvidia’s China challenge may be turning structural
The piece notes that NVIDIA Corporation (NASDAQ:NVDA) still has the stronger global platform, supported by CUDA, networking and the pace of its system-level launches. But export restrictions have already limited Nvidia’s ability to serve the highest-end segment of China’s demand.
That means Alibaba’s gains may initially replace sales Nvidia is largely unable to make under current rules anyway. The longer-term risk, however, is that local customers are being forced to optimize around domestic hardware, turning a temporary supply constraint into lasting ecosystem development inside China.
According to Insider Monkey, 97 hedge funds reported long positions in Alibaba in the second quarter of 2026, down from 102 in the first quarter. Fisher Asset Management held about 5.10 million shares after trimming its stake by 1%. For Nvidia, hedge-fund holders rose to 285 from 275, while Fisher increased its Nvidia position by about 3%. Those filings, however, came before the V900 launch.
As of August 14, Alibaba short interest stood at about 41.98 million shares, or roughly 2.0% of public float, with around 4.6 days to cover.
The article’s conclusion is that Alibaba does not need to beat Nvidia across the board for the development to matter at the stock level. It needs a chip good enough to handle a growing share of Chinese AI workloads. Each generation that narrows the gap reduces the chances that Nvidia could simply reclaim its old position in China if policy conditions change later.
