Skip to content
Advertisement
Alibaba executive signals its 20 GW data center target may shrink

Alibaba executive signals its 20 GW data center target may shrink

Shifting global compute policy could force Alibaba to scale back its cloud ambitions while it leans harder into domestic chips

Two months ago, Alibaba set one of the most aggressive infrastructure goals in the cloud industry. Now one of its own executives is suggesting the number might need to come down.

An Alibaba executive indicated on November 19, 2026, that changes in global compute policy may lead the company to revise its 20 gigawatt data center capacity target downward. The company also expects major investment in domestic chips to arrive in 2027 and 2028.

From 20 GW ambition to a possible rethink

The original goal was announced by Alibaba CEO Eddie Wu on September 22, 2026, at the company’s Apsara Conference. The plan called for Alibaba Cloud to exceed 20 GW of global data center capacity by 2032.

That announcement was framed as a signal of strong industry demand. It was also tied to Alibaba’s in-house silicon from its T-Head chip unit.

The headline chip in that lineup is the Zhenwu V900 AI processor. Alibaba says it delivers three times the performance of its predecessor.

Wall Street took the target seriously. Analysts at Goldman Sachs, Citi, and UBS modeled capital expenditures of approximately 200 to 260 billion RMB per year following the 20 GW announcement.

Advertisement

Those same models pointed to a big payoff. External cloud revenue tied to the expansion was estimated at between $160 billion and $170 billion by 2032 to 2033.

Washington changes the math

The timing is hard to miss. Just two days before the executive’s remarks, on November 17, 2026, draft US regulations were reported that target loopholes in chip access for Chinese firms.

Specifically, the draft rules take aim at Chinese hyperscalers reaching advanced semiconductor technology through offshore data centers.

According to the research summary, the US policy developments have introduced uncertainty. Analysts have shifted toward a more cautious stance on the growth trajectories of Chinese AI and cloud services.

Nothing has been formally cut yet. The executive’s language was that the target may be revised, and the US rules remain in draft form.

The domestic chip pivot

The other half of the executive’s message points in a different direction. Alibaba expects significant investment in domestic chips, with that investment phase projected to peak in 2027 and 2028.

The Zhenwu V900 is the clearest example of where Alibaba wants to go. A threefold performance jump over its predecessor is the kind of claim that makes a self-reliance strategy sound more credible.

What this means

For investors watching Chinese tech, the immediate question is how much of that 200 to 260 billion RMB annual capex modeling survives. A smaller capacity target could mean lower spending, which some shareholders might welcome in the short term.

The flip side is the revenue projection. The $160 billion to $170 billion external cloud revenue estimate was built on the original buildout. If capacity shrinks, the analysts who produced those figures may need to revisit them.

The draft US rules apply to Chinese hyperscalers as a group. Alibaba is simply the first to say out loud that its plans could change as a result.

The key things to watch are straightforward. First, whether the draft US regulations are finalized and in what form. Second, whether Alibaba puts a new number on its capacity goal or keeps the 20 GW figure with softer language. Third, how quickly domestic chips like the Zhenwu V900 can scale to fill the gap.

Disclosure: This article was edited by John Chen. For more information on how we create and review content, see our Editorial Policy.