Huawei Unveils Chip and Computing Power Roadmap in Direct Challenge to Nvidia

Huawei has laid out its long-term semiconductor and computing strategy, signaling China’s determination to reduce reliance on foreign suppliers and intensify competition with U.S. chip giant Nvidia. The announcement, made at the Huawei Connect conference in Shanghai, represents the company’s most detailed disclosure yet about its chip ambitions since U.S. sanctions in 2019 forced it to scale back. 

Eric Xu, Huawei’s rotating chairman, said the company will pursue a one-year release cycle for its Ascend artificial intelligence chips, doubling computing power with each iteration. The roadmap includes the launch of the Ascend 950 next year, followed by the 960 in 2027 and the 970 in 2028. Xu also revealed that Huawei has developed proprietary high-bandwidth memory, a space long dominated by South Korea’s SK Hynix and Samsung Electronics. 

The company’s push extends beyond chips to what it calls “supernodes”—large-scale computing clusters that interconnect thousands of chips. Huawei expects to launch the Atlas 950 in late 2026, supporting 8,192 Ascend chips, followed by the Atlas 960 in late 2027, which will support 15,488 Ascend chips. The firm claims these supernodes will far exceed international rivals across key performance benchmarks. 

Raising the Stakes in the U.S.–China Tech Rivalry

Huawei’s timing is notable. The announcement comes just as Chinese authorities accused Nvidia of violating antitrust rules, and China’s internet regulator reportedly ordered major tech companies to suspend purchases of its AI chips and cancel existing orders, according to a Financial Times report. Meanwhile, U.S. President Donald Trump and Chinese President Xi Jinping are preparing to meet this Friday following fresh rounds of trade negotiations, underscoring how technology has become central to geopolitical tensions. 

Analysts suggest the rollout is intended to project confidence in China’s domestic capabilities. “China is trying to show progress across multiple fronts,” said Alfred Wu, associate professor at the National University of Singapore. “But the reality is that tensions with the U.S. are quietly escalating, not easing.” 

Huawei began exploring chip design in 2018 but retreated from public discussion after sanctions cut off access to critical U.S. technology. Since then, the company has emerged as a leading force in China’s effort to build a self-sufficient semiconductor ecosystem. Alongside its AI chips, Huawei is also planning upgrades to its Kunpeng server processors in 2026 and 2028. 

Industry experts note that Huawei is leveraging its strengths in networking and China’s relatively stable power supply to advance large-scale computing clusters, even as domestic chip manufacturing lags behind top global players. News of Beijing’s restrictions on Nvidia chip purchases lifted Chinese semiconductor stocks by about 2% in afternoon trade, reflecting investor confidence in a shift toward homegrown alternatives. 

Summary:

Huawei has unveiled a multi-year roadmap for AI chips and high-performance computing clusters, positioning itself as China’s answer to Nvidia. With new Ascend chips, supernodes, and proprietary memory technology in development, the company is signaling both technological ambition and geopolitical resolve. The move could reshape competition in the AI hardware market while intensifying U.S.–China tech tensions.

Frequently Asked Questions (FAQs)

1. What are Huawei’s Ascend chips?

Ascend chips are Huawei’s line of artificial intelligence (AI) processors designed for high-performance computing and data-intensive applications. The company plans yearly upgrades that will double computing power with each release.

2. What are Huawei’s “supernodes”?

Supernodes are large-scale computing clusters that connect thousands of AI chips to work together at high speed. Huawei’s upcoming Atlas 950 and Atlas 960 systems are designed to outperform existing global competitors in terms of processing capacity.

3. Why is Huawei developing its own chips?

Huawei accelerated its chipmaking efforts after U.S. sanctions in 2019 cut off access to critical American technology. By investing in its own semiconductor ecosystem, the company aims to reduce reliance on foreign suppliers like Nvidia and strengthen China’s domestic tech independence.

4. How does Huawei’s chip strategy challenge Nvidia?

Nvidia dominates the global AI chip market. Huawei’s roadmap—including proprietary memory technology and powerful computing clusters—directly positions it as a rival, especially in markets where access to Nvidia products may be restricted.

5. What role does high-bandwidth memory play?

High-bandwidth memory (HBM) improves data transfer speeds between processors and memory. Huawei’s move to develop its own HBM reduces dependence on SK Hynix and Samsung, which currently lead in this field.

6. What are the geopolitical implications of Huawei’s announcement?

The timing of Huawei’s roadmap coincides with escalating U.S.–China tensions over technology. China recently restricted purchases of Nvidia’s AI chips, signaling a push toward homegrown alternatives. Huawei’s plans showcase China’s determination to build its own semiconductor supply chain.

7. When will Huawei’s new chips and supernodes be available?

Ascend 950: 2026

Ascend 960: 2027

Ascend 970: 2028

Atlas 950 supernode: late 2026

Atlas 960 supernode: late 2027
Upgrades to Kunpeng server chips are also planned for 2026 and 2028.

8. How has the market responded so far?

Shares of Chinese semiconductor firms rose about 2% after reports that Beijing ordered domestic tech companies to halt Nvidia chip purchases. This suggests investors are optimistic about China’s shift toward local alternatives like Huawei.