China’s New Dream: On the Path to Building Its Own Nvidia

Dr. Mashiur Rahman

A new chapter in the global technology race is being written at one end of Asia. As the conflict with the United States intensifies, China is increasingly realizing that simply dazzling the world with generative AI models is not enough. To run these models, chips are needed—and reducing foreign control over those chips is now crucial. Beijing is determined to break Nvidia’s long-standing dominance, a pursuit significant not only in economic terms but also as a matter of national strategy.

A major opportunity has suddenly arisen. News has emerged that Nvidia has halted production of its H20 AI chip, which was custom-made for China. Citing national security, Beijing had already advised technology companies to refrain from purchasing that chip. Soon after, Chinese firm DeepSec unveiled an upgrade to its flagship V3 model, with provisions for a new generation of domestic chips. This is no mere coincidence; rather, it signals that China has now entered the battle to run its own AI on homegrown hardware.

So far, Nvidia still leads the global market. The high-performance chips required for training generative AI models are largely in their hands. But technological evolution brings a new reality—for generating answers from pre-trained models or inference tasks, the necessary chips are much less complex and more affordable. U.S. Commerce Secretary Howard Lutnick disparagingly remarked that the H20 chip “doesn’t even rank among our top three,” yet that chip was selling in the Chinese market and was well-suited to such tasks. It’s not hard to imagine that, over time, Chinese companies designing application-specific integrated circuits (ASICs) can quickly close the gap with Nvidia.

However, the challenges are considerable. For many years, China has poured government funding into creating “national champions.” But the results have not always met expectations. The bankruptcy of Tsinghua Unigroup in 2021 and subsequent corruption investigations into the China Integrated Circuit Industry Investment Fund, known as the “Big Fund,” are glaring examples of such failures. Excessive government support often breeds waste, debt, and a web of complex liabilities.

In this context, the rise of Cambricon Technologies shines as a beacon. Founded in 2016, this AI chip designer was listed on the Shanghai Stock Exchange’s STAR Board in just four years—a pilot program launched by President Xi Jinping to fast-track technology startups onto the stock market. Investors see such promise in China’s prospects in the inference chip market that Cambricon is now planning to raise nearly 4 billion yuan (about 560 million US dollars) for research and development. Sales are rising, yet this young company is still operating at a loss—but confidence in the market is propelling them forward.

There is a lesson here. Complex structures of state patronage, as seen with private companies like Unigroup, have ultimately led only to confusion and bankruptcy. In contrast, the transparency of the stock market and the natural due diligence of investors are far more effective. Here, the government primarily acts as a policy setter, but market forces themselves evaluate actual risks and opportunities.

This shift can also be seen in China’s recent policies. As Tilly Zhang from research institute Gavekal Dragonomics puts it, the government is now leaning toward an “early, small, hard” investment policy—encouraging venture capital to support the core technologies of small companies. Cambricon’s founder Chen Tianshi still owns 29% of the company’s shares, while an affiliate of the Chinese Academy of Sciences holds 16%. The combination of private ownership and research institutions reflects China’s new direction.

At the same time, President Xi’s clear warning to local governments in less prosperous regions against making excessive investments in hard technology proves that Beijing is now choosing a more discerning path instead of simply pouring in money.

No matter how long and hard this road may be, the direction is clear. If China truly wants to build an alternative to Nvidia, the key will be to encourage young startups to enter the stock market as quickly as possible. Because as investors search for their own “Chinese Nvidia,” they themselves will perform the necessary vetting on behalf of the government. Out of this process will rise genuine national champions—bringing China’s dream of technological self-reliance ever closer to reality.