Invest1 publisher3 min readPublished
Huawei's faster AI chip timetable runs into its chairman's warning on supply
Huawei moved its next AI chip three quarters forward to a first-quarter 2027 ship date, days before Xi Jinping's first White House visit in 11 years. Its chairman says Huawei may not make enough chips for domestic demand, so output remains the limit on Chinese AI compute.
The Investor · Invest desk

What happened
- Alibaba's chip unit unveiled the Zhenwu V900, which Alibaba called China's most powerful AI processor, with triple its predecessor's performance and sales from early 2027.
- The launches came days before Xi Jinping's first White House visit in 11 years, a trip that follows Donald Trump's state visit to Beijing in May.
- Scott Bessent and Jamieson Greer met Vice Premier He Lifeng in New York over the weekend and set up a formal AI channel for incident warning.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- constraint Chinese developers building on Huawei need about two chips per Nvidia equivalent, so Huawei's output sets their pace more than its launch calendar does.
- cost Chinese AI firms pay for that larger chip count out of revenue that is a small fraction of their US rivals', so the controls bite through budgets as well as through access.
- contradiction Wang sees genuine production acceleration at Huawei while Lu sees no breakthrough in years, and Huawei's own supply warning is the only volume evidence either view can point to.
A first-quarter 2027 ship date that is three quarters early means Huawei originally planned the 960DT for the fourth quarter of 2027 [7]. Both flagship chips on show this week are products for next year [1][2]. Chris Miller, the Tufts University professor who wrote "Chip War", said China has a history of announcing new projects around meetings with senior policymakers, as part of an effort to show its firms are "catching up" [8].
Washington's bet, made over years of restricting Nvidia's most advanced chips and the equipment to make them, was that starving China of computing power would keep its AI a generation behind [5]. For Xi, said Lizzi Lee, a fellow at the Asia Society Policy Institute, the leverage is not "we have caught up with you" but "you can't simply technologically choke us off" [9].
The slowing that Lee's line concedes shows up first in compute. Each Huawei chip delivers roughly half the compute of Nvidia's [10], so matching an Nvidia installation takes about two Huawei chips for each one [11]. Chinese firms have tried to close that gap by linking thousands of chips into clusters over faster connections [12]. Huawei's latest SuperPoD architecture links far fewer processors than originally planned [13]. Doubling the chip count also requires the chips to exist, and chairman Eric Xu acknowledged that Huawei may not be able to make enough of them to meet domestic demand [14].
It shows up second in cash. Miller said American firms such as Anthropic and OpenAI make 50 to 100 times the revenue of comparable Chinese firms [15]. Divide one by those multiples and the Chinese firm earns between 1% and 2% of its American peer's revenue [16], while buying roughly twice as many chips for the same compute [11]. Miller said Chinese firms rely heavily on smuggled chips or access to data centers outside China [17]. "China is still highly dependent on the U.S. for advanced chips," he said [18].
The case against that reading starts with the schedule itself. "There does seem to be genuine acceleration in its production timeline," Sigrid Wang, an analyst at Hutong Research, said of Huawei's chip [19]. DeepSeek has reportedly lined up the largest known cluster of Huawei chips built to date [3]. "Chinese chip companies haven't made any groundbreaking progress in the past few years," said Xiaomeng Lu, a director at Eurasia Group [20].
The earlier date fits more than one reading. It could mean more output, in which case Wang is right and the controls buy Washington less time than it bet. It could be the same constrained output arriving sooner. Or it could be summit timing of the kind Miller described, with a date that later slips. I think the second is best supported, because the only statement on volume comes from Huawei's chairman and it points to a shortfall [14]. That view is wrong if Huawei says it can meet domestic demand for the 960DT. For now, neither side appears ready to put chip export controls on the table [21].
What to watch
- Any change to US restrictions on Nvidia chips or chipmaking equipment in the weeks after Xi's multi-day visit ends.
- Confirmation of the size of DeepSeek's reported Huawei cluster, and whether a model trained on it is released.
- Whether Alibaba's Zhenwu V900 reaches buyers in early 2027 at the tripled performance the company claimed.