China’s ‘homegrown Nvidia’ chief launches new AI chips

Moore Threads is back in the spotlight after its wild Shanghai debut, and Zhang Jianzhong walked right into that momentum with new AI chips meant to cut dependence on Nvidia inside China.

The company rolled out this new hardware only weeks after its stock jumped five-fold, and Zhang used the Beijing event to push a clear message: developers in China should not have to wait for foreign tech to move forward.

“These products will significantly enhance world-class computing speed and capabilities that all developers aspire to,” he said, adding that he wants local teams to build without waiting for outside supply. “We hope they can meet the needs of more developers in China so that you no longer need to wait for advanced foreign products.”

China’s chipmakers are getting heavy attention as the government pushes for a strong semiconductor base, and investors are betting that firms like Moore Threads can stand up to Nvidia at a time when US export bans block the most advanced chips from reaching China.

Zhang said the new Huagang architecture boosts computing density by 50% and lifts energy efficiency by ten times, giving local developers more room to work with large models without depending on outside hardware.

Chips made on this design will fall under the Huashan line and target the same category as Nvidia’s Hopper and Blackwell units.

Building hardware shifts across China’s chip sector

Moore Threads went public in Shanghai literally this month and saw its shares jump more than five-fold, followed by domestic rival MetaX Integrated Circuits posting a strong debut days later.

Zhang founded Moore Threads back in 2020 after 14 years at Nvidia, where he served as China general manager and helped build the local ecosystem he now wants to replace.

Before Nvidia, Zhang’s path took him through roles at Hewlett-Packard and Dell, and before that, he started as a senior researcher at the Metallurgy Automation Research & Design Institute in 1990.

The company said it expects mass production in 2026, with the new technology able to support clusters of over 100,000 chips inside data centers for AI training. Moore Threads first made money from gaming and visual rendering chips, and later switched to AI accelerators as demand for local options grew.

At the same event, the company rolled out an update to its MUSA computing platform, calling it an equivalent to CUDA, and introduced servers that can link tens of thousands of AI chips, even after being blacklisted by the US in 2023.

Expanding product lines across local hardware

Moore Threads also unveiled the Lushan GPU series for graphics rendering and launched the Changjiang SoC, which puts CPUs and GPUs on the same chip.

Analysts watching China’s market say the strong share moves may come from excitement rather than fundamentals.

“In the context of the US-China tech war, valuations in the A-share market have detached from reality, driven by politics rather than logic,” said Shen Meng of Chanson & Co. He added that many of these companies act as political symbols and show limited real impact on core technology.

Zhang said “policy support is the ‘booster’ for strategic high-tech breakthroughs,” pointing to the capital-heavy nature of chips. The company’s filings show Moore Threads expects to turn a profit by 2027.

Duncan Clark, Chairman of BDA China, said this is a “domestic substitution play,” noting that China’s push for local chips gives Moore Threads a clear path since government-backed buyers must source hardware domestically, including sectors like the military.

Claim your free seat in an exclusive crypto trading community - limited to 1,000 members.