Nvidia CEO Jensen Huang downplayed US concerns about the use of the company’s chips in China’s military ahead of another visit to the country.
He said China’s military would not use American-made technologies because “they simply cannot rely on them.”
“Their access can be restricted at any moment, not to mention that China already has sufficient computing power. They do not need Nvidia chips, and certainly do not need American technology stacks, to develop their armed forces,” Huang said.
The US has for years imposed restrictions on semiconductor makers, barring them from selling the most advanced AI processors to China.
The latest curbs were introduced in April, when Nvidia was forced to obtain a licence to export H20 chips. In response, the company began developing an AI chipset for the Chinese market at a significantly lower price, according to media reports.
In the first quarter, Huang’s company incurred expenses of $4.5bn due to excess H20 inventories and missed out on $2.5bn in potential additional sales. Despite the “effective closure of the China market,” the firm became the first in history to reach a $4trn market capitalisation.
Nvidia’s boss again criticised the policy, stressing that export controls backfire against the very goal of maintaining US leadership in technology.
“We want the American technology stack to become the global standard. To achieve that, we need to work with all AI developers in the world. Half of them are in China,” he said.
Huang’s interview was published a few days before his second trip to China this year. On July 10 he met President Donald Trump at the White House. At the same time, US lawmakers warned the executive against meetings with companies linked to China’s military or intelligence services, as well as organisations on the restricted export list.
DeepSeek aids China’s military
AI startup DeepSeek is helping China’s military and intelligence operations, reported Reuters, citing a senior American official.
“We understand that DeepSeek has willingly provided and will likely continue to provide support to China’s military and intelligence operations. These efforts go beyond open access to the company’s artificial intelligence models,” the State Department source said.
According to the agency, the firm tried to use shell companies in Southeast Asia to gain access to high-tech, restricted semiconductors.
“It is trying to gain access to data centers in the region for remote access to American chips,” the source said.
In Washington, there is growing confidence that the rapidly growing Chinese startup’s capabilities are overstated and rely heavily on American technologies.
“We do not support parties that have violated US export controls or are on US lists. With the current export controls, we have effectively exited the Chinese data center market, which is now served only by competitors like Huawei,” Nvidia commented.
Under Chinese law, companies operating in the country must provide data to the government upon request. US authorities have previously said that DeepSeek transfers American users’ personal information to China through “internal infrastructure” linked to the telecom provider China Mobile.
The startup is also mentioned more than 150 times in procurement records for the People’s Liberation Army and other organisations connected to the defence industrial base, Reuters writes.
At the same time, the firm uses workarounds to US export controls to obtain advanced chips. It has access to “large volumes” of H100 processors, which have been restricted since 2022.
AI is central to China’s military
China views AI as a key factor in future conflicts and is seeking to shift to “intelligentised” warfare. The main areas include:
- autonomous weapons and unmanned systems — priority is given to developing independent combat platforms; “drone-swarm” technologies are being explored for coordinated massed attacks and the use of lethal autonomous weapons;
- intelligence, surveillance and targeting — AI is used to automatically process vast volumes of intelligence and surveillance data; machine-learning algorithms are used to recognise targets in imagery and rapidly identify threats;
- cyber operations and information warfare — algorithms help defend domestic networks and attack adversary systems, from filtering network traffic to creating countermeasures;
- command and communications systems — AI is integrated into the processes of command, control, communications and intelligence to support decision-making.
In China, the development of military AI proceeds through close cooperation between the state and the private sector under the personal oversight of the country’s leadership. Key players include:
- Norinco — the largest state defence conglomerate, a manufacturer of arms and military equipment. Norinco invests in robotics and autonomous platforms. It unveiled the Intelligent Precision Strike System — a system that automatically coordinates groups of strike drones, models the battlefield, tracks targets and plans strikes with minimal human involvement;
- China Electronics Technology Group Corporation (CETC) — a leading state electronics corporation responsible for radars, communications systems and electronics for the PLA. CETC is actively engaged in military AI technologies — from electronic warfare tools to unmanned systems;
- Baidu — one of the largest private tech giants, heavily involved in AI programmes. It focuses on autonomous driving technologies, big data and cloud AI. The company collaborates with the defence sector via joint projects (for example, a lab with CETC) and provides its research for military applications within the policy of civil–military fusion;
- SenseTime — a major private firm and a leader in computer vision and AI-powered image analysis. It specialises in facial recognition and video analytics. Its technologies are widely used in public-security systems, and the know-how is assumed to be applicable to military purposes — from range surveillance to intelligence analysis;
- iFLYTEK — a private company specialising in speech recognition and language AI. It is one of the “national champions.” The firm’s products are used by the PLA: speech analysis and synthesis modules help automate monitoring of radio communications and interception for intelligence tasks.
Financing
Data on the financing of military AI programmes in China are partly classified, but available estimates suggest the scale of investment is considerable.
A study by the Center for Security and Emerging Technology of open PLA contracts found a minimum level of around $1.6bn a year for the procurement of systems with AI elements. Actual spending is likely higher, as a significant share of investment is hidden within R&D and classified programmes.
China and the US are effectively engaged in an arms race in artificial intelligence. The Chinese military is rapidly deploying AI across a broad range of applications — from unmanned platforms to data analysis — in an effort to catch up with or surpass the US. America still leads in several critical technologies and is taking steps to retain that leadership.
Both countries are investing heavily in military AI, recognising that mastery of these technologies will largely determine the balance of power and the nature of security in the 21st century.
In a remote desert area of the north-western Xinjiang region, China is building dozens of data centers to house processors. Companies from China plan to purchase more than 115,000 of Nvidia’s high-end chips to power the new capacity and further train AI models.