Stop overpaying - start transferring money with Ogvio. Sign up, invite friends & grab Rewards now! 🎁
Huawei Ramps Up AI Chip Race with New Ascend 910D Launch
Key Takeaways
- Huawei will begin testing its new Ascend 910D AI chip by May, with the aim of competing with Nvidia’s H100 chips;
- Over 800,000 Ascend 910B and 910C chips are being delivered to telecom firms and companies like ByteDance;
- China is pushing developers to adopt local chips as trade tensions with the United States continue to rise.
Huawei, a Chinese multinational technology company, is planning to test its latest artificial intelligence (AI) chip, the Ascend 910D.
According to sources cited by The Wall Street Journal, the company expects to deliver the first samples to local technology firms by the end of May. These companies will run early tests to check the chip’s capabilities before it becomes available for broader use.
The Ascend 910D is being developed to compete with Nvidia’s H100 chip, which has been widely used for AI model training since 2022. Huawei hopes its new chip will offer stronger performance.
Did you know?
Subscribe - We publish new crypto explainer videos every week!
Crypto Research Fundamentals: How to DYOR (Animated Explainer)
While the new chip is still being prepared, Huawei is also distributing over 800,000 units of its earlier models—the Ascend 910B and 910C. These chips are being supplied to a range of customers, including state-owned telecom providers and private companies such as ByteDance, the parent company of TikTok.
Chinese leaders are encouraging local developers to use domestic chips, especially as trade tensions between China and the United States continue to increase. Companies are being encouraged to use local suppliers through a combination of policy support and public backing.
At the same time, Huawei is working on building larger computing systems that use many chips together. In April, the company introduced CloudMatrix 384, a system that links hundreds of Ascend 910C chips. Rather than focusing on improving individual chips, this approach combines the power of many processors to boost performance.
Meanwhile, on April 14, Nvidia announced plans to invest $500 billion to build AI chip and supercomputer facilities in Phoenix, Dallas, and Houston. What did CEO Jensen Huang say about it? Read the full story.