BEIJING — A Chinese university with close ties to the country’s military has reportedly obtained some of Nvidia’s most powerful artificial intelligence processors despite being placed on a United States trade blacklist, raising fresh questions about whether Washington can effectively prevent restricted technology from reaching sensitive institutions.
According to procurement records reviewed by Reuters, Northwestern Polytechnical University purchased two Nvidia H200 AI chips in October for 874,700 yuan (approximately US$123,000 or S$168,158).
The purchase was recorded on October 21 and made through a domestic Chinese supplier, Shaanxi Bosuan Network Technology, rather than directly from Nvidia.
The university is particularly sensitive from Washington’s perspective. It belongs to China’s so-called Seven Sons of National Defence group, a collection of universities with extensive defence-related research connections.
The United States added Northwestern Polytechnical to its Entity List in 2001 over alleged involvement in missile-proliferation activities and military research. Restrictions were tightened in 2019, with export licences for controlled American technology generally subject to a policy of denial.
But the reported purchase is only part of a much larger picture.
Reuters examined 26 Chinese procurement and corporate records and found that universities and research institutions had obtained restricted Nvidia processors or rented computing capacity powered by them through domestic suppliers.
Some of the institutions have connections to China’s defence and military research sectors.
The findings raise an uncomfortable question for Washington: If export controls are designed to restrict access to advanced AI technology, how are institutions on the US blacklist still obtaining access to powerful Nvidia computing hardware?
Two Nvidia H200 Chips Purchased for Approximately US$123,000
The procurement record identifies Northwestern Polytechnical University as the buyer of two Nvidia H200 processors.
The chips were purchased for the university’s management school from Shaanxi Bosuan Network Technology.
The transaction was valued at 874,700 yuan, equivalent to approximately US$123,000.
The purchase is significant because the H200 is among Nvidia’s most capable AI processors, designed to support demanding workloads such as training and operating advanced AI models.
Unlike ordinary consumer graphics cards, high-performance AI accelerators are valuable components of large computing systems used by technology companies, research institutions and governments.
Their capabilities make them strategically important in the global competition over artificial intelligence.
However, the procurement record does not establish where the two processors were originally manufactured, how they entered China, or whether they were exported in violation of US law.
That distinction matters.
The reported purchase demonstrates that the university obtained the chips through a Chinese supplier. It does not, by itself, establish that Nvidia directly sold the processors to the university or that a specific illegal export transaction occurred.
Reuters reported that it could not determine the origin of the chips or obtain comment from the supplier.
Northwestern Polytechnical, Nvidia and the US Commerce Department’s Bureau of Industry and Security did not respond to Reuters’ requests for comment.
Why Northwestern Polytechnical University Is on Washington’s Blacklist
Northwestern Polytechnical University, based in Xi’an in northwestern China, is known for research involving aerospace engineering, military aircraft, missiles and satellite technology.
It is also a member of the Seven Sons of National Defence, a group of Chinese universities with significant defence-related research activities.
Washington added the institution to its Entity List in 2001, citing alleged involvement in missile-proliferation activities and military research.
In 2019, US authorities tightened restrictions, making licences for exports of controlled American technology to the university subject to a policy of denial in most circumstances.
The Entity List is administered by the US Department of Commerce.
It identifies foreign organisations and individuals considered to present national-security or foreign-policy concerns. Transactions involving controlled US-origin items and listed entities generally require a licence.
The restrictions are particularly relevant to advanced AI processors because Washington considers certain high-performance computing technologies capable of supporting military modernisation and other sensitive applications.
However, being listed does not mean every commercial transaction involving a university is automatically illegal under every jurisdiction. The specific product, export route, transaction and applicable licensing requirements matter.
The key issue in the Northwestern Polytechnical case is whether the processors or associated computing services were obtained through arrangements that complied with the relevant export-control rules.
That question remains unresolved in the publicly available procurement records.
The Bigger Discovery: The University Also Rented Access to Nvidia AI Chips
The two H200 processors were not the university’s only reported access to advanced Nvidia technology.
Reuters found that Northwestern Polytechnical also obtained computing capacity through five contracts with Chinese suppliers during 2026.
The arrangements involved access to servers equipped with Nvidia H200, A100 and A800 processors.
Among the records were two seven-month contracts awarded by the university’s cybersecurity school in January.
Each contract covered access to servers containing eight H200 chips.
The contracts were valued at approximately 441,700 yuan and 441,000 yuan.
Their stated purposes included fine-tuning AI models and reinforcement-learning training.
The university also rented access to servers containing eight Nvidia A100 chips and eight A800 chips.
The A100 and A800 processors are also subject to US export restrictions for China.
The reported arrangements raise a particularly important issue for export enforcement: restricting the physical sale of a processor may not be enough if an organisation can obtain access to computing power through a domestic cloud or server provider.
A customer does not necessarily need to own an AI processor to use its computing capabilities.
Cloud-computing arrangements can provide remote access to powerful hardware, potentially allowing organisations to perform advanced AI workloads without directly purchasing the equipment.
That makes the location of the hardware, the identity of the customer and the nature of remote access important considerations for regulators.
US Export Rules Changed in January — But the Blacklist Restrictions Remained
The situation is complicated by changes in US policy.
In January 2026, President Donald Trump’s administration began allowing certain Nvidia H200 exports to China on a case-by-case basis.
The policy represented a shift from earlier restrictions that had sharply limited China’s access to advanced American AI processors.
However, the January policy did not remove the separate restrictions applying to organisations on the Entity List, including Northwestern Polytechnical University.
The licensing framework also included safeguards intended to prevent blacklisted entities and military end-users from remotely accessing chips exported under the approved arrangements.
This distinction is central to understanding the latest procurement findings.
The United States did not simply open the door for every Chinese institution to purchase H200 processors.
Some exports could be considered under licensing conditions, but restricted entities remained subject to separate controls.
Consequently, the reported university purchases raise questions about the effectiveness of monitoring, enforcement and the ability to identify the ultimate users of advanced computing hardware.
The findings do not establish that the January policy authorised Northwestern Polytechnical to receive the processors.
Reuters Finds a Wider Network of Restricted Nvidia Chip Access in China
The Northwestern Polytechnical case is not an isolated procurement record.
Reuters reviewed 26 Chinese procurement and corporate records from 2026 and identified other universities and public institutions purchasing restricted Nvidia processors or renting access to computing services powered by them.
The records included purchases involving H100, A100, A800 and H800 processors.
Examples included:
- Shenzhen University: Purchased three Nvidia H100 processors for 769,800 yuan in September.
- Wuhan University of Technology: Purchased a server containing an H100 processor for 355,000 yuan in June and four A100 processors in a separate transaction valued at 344,000 yuan.
- Beihang University: Purchased four Nvidia A800 processors for 489,000 yuan in April.
- University of Chinese Academy of Sciences: Procurement announcements identified suppliers offering computing services using Nvidia H800 and A800 processors.
- A Shenzhen-based public institution: A January procurement award identified a supplier of six A100 processors valued at 484,800 yuan.
Some of the institutions have connections to China’s aerospace, defence or military research sectors.
The records also revealed that several purchases were made through relatively little-known domestic technology suppliers.
However, the documents often provided limited information about where the processors were physically located, where the servers operated or how the hardware originally entered the country.
That information gap is important.
A procurement contract showing access to restricted technology does not automatically establish how the hardware was sourced or whether an export-control violation occurred.
But the findings demonstrate why regulators are increasingly concerned about indirect access through intermediaries and computing-service arrangements.
Why the H200 Matters in the Global AI Race
Nvidia’s H200 is a high-performance graphics processing unit designed for demanding artificial intelligence and high-performance computing workloads.
It uses high-bandwidth memory and is built to support large AI models that require substantial computing resources.
AI processors are essential for training sophisticated models, processing enormous datasets and operating advanced AI services.
They are also relevant to scientific simulations, engineering research and other computationally intensive applications.
The same capabilities can have military implications.
Advanced AI systems can potentially assist with intelligence analysis, autonomous technologies, military logistics, surveillance and research connected to weapons development.
This is why Washington has treated certain AI processors as strategic technology rather than ordinary commercial products.
The concern is not that every AI processor purchased by a Chinese university will automatically be used for military purposes.
Rather, US policymakers want to prevent sensitive computing capabilities from reaching organisations or end-users considered capable of supporting China’s military and strategic objectives.
The Northwestern Polytechnical procurement records have brought that debate back into focus.
The Underground Market and Third-Country Routes Complicate Enforcement
Reuters has previously documented how restricted Nvidia processors have reached Chinese customers through indirect supply arrangements.
In some cases, advanced chips were reportedly embedded in servers manufactured by companies such as Super Micro Computer, Dell and Gigabyte before being acquired through domestic resellers.
The agency has also reported on an underground market involving restricted processors sourced from overseas inventories and third-country intermediaries.
These arrangements illustrate the difficulty of monitoring technology once it enters a complicated international supply chain.
A processor can move through manufacturers, distributors, server integrators, cloud providers and resellers before reaching its eventual user.
That makes it difficult for regulators to establish the final destination and end-user without reliable documentation and effective monitoring.
The problem extends beyond China.
Taiwanese authorities have also investigated alleged attempts to divert high-performance AI servers containing restricted technology to China.
As WWC ONE MEDIA reported in Taiwan Indicts 9 in AI Server Export Case — But Where Were the Powerful Machines Really Headed?, prosecutors accused nine individuals of involvement in a separate alleged scheme involving high-performance AI servers and restricted technology.
That Taiwanese case is separate from the Northwestern Polytechnical procurement findings. It does not establish that the same suppliers or individuals were involved.
Nevertheless, both developments highlight the challenge of controlling advanced AI hardware across international supply chains.
Nvidia CEO Jensen Huang Faces Growing Political Scrutiny
The latest revelations also arrive at a sensitive time for Nvidia and its chief executive, Jensen Huang.
Huang has argued that allowing American technology companies to compete in international markets helps preserve US leadership in artificial intelligence.
His position is that excessive export restrictions could encourage Chinese customers to adopt domestic alternatives, potentially strengthening competitors such as Huawei over time.
However, some US lawmakers believe allowing China access to powerful American AI processors could accelerate its military and technological capabilities.
Both Democrats and Republicans have called for stronger oversight of advanced chip exports.
Three Republican congressional committee chairmen have requested a briefing with Huang on November 18 over his reported contacts with an alleged Chinese intelligence operative.
The controversy has increased scrutiny of Nvidia’s lobbying for looser restrictions on sales to China.
It is important, however, to distinguish the separate issues.
The university’s procurement records do not establish that Huang or Nvidia authorised the reported purchases. Nor do they prove that the company knowingly supplied restricted processors to a blacklisted institution.
The latest findings instead raise questions about the effectiveness of the wider export-control system.
WWC ONE MEDIA previously examined the political implications in US Congress Demands Answers From Nvidia as Jensen Huang Spy Allegations Rock Washington — But Could China Lose Access to America’s AI Chips?.
The congressional scrutiny reflects a broader disagreement over whether Washington should prioritise commercial competition or impose tighter restrictions to protect sensitive technology.
China Is Also Trying to Reduce Its Dependence on Nvidia
There is another important dimension to the dispute.
China is not simply trying to obtain American AI processors. It is also investing heavily in its own semiconductor industry.
Domestic companies, including Huawei, are developing processors and computing systems intended to reduce reliance on foreign technology.
The country’s AI industry needs enormous computing capacity to train and operate increasingly sophisticated models.
Access to Nvidia hardware could provide an important advantage, but Chinese policymakers also have an incentive to strengthen domestic alternatives.
That creates a complicated commercial and political situation for Nvidia.
The company wants to preserve access to one of the world’s largest AI markets, while Washington wants to restrict technology transfers that could undermine American national-security interests.
Beijing, meanwhile, must balance the immediate demand for powerful computing hardware against its longer-term ambition to build a more self-reliant semiconductor industry.
WWC ONE MEDIA previously explored that tension in China Weighs Allowing ByteDance and Alibaba to Buy New Nvidia Chips as AI Demand Surges.
The latest Northwestern Polytechnical findings show why the debate is becoming increasingly difficult to resolve.
Even when direct sales are restricted, demand for computing power can create alternative routes through suppliers and service providers.
What the Procurement Records Do Not Prove
Despite the significance of the findings, several important questions remain unanswered.
First, where did the two H200 processors originate?
The procurement record identifies the domestic supplier but does not establish the processors’ original source or route into China.
Second, where were the rented servers physically located?
The contracts identify access to computing capacity, but the records reviewed by Reuters offered limited information about server locations.
Third, did any of the transactions violate US export-control laws?
The available records raise compliance questions but do not independently establish a specific unlawful export or identify the party responsible.
Fourth, were the processors used for military research?
Northwestern Polytechnical has defence-related research activities, but the procurement records do not prove that the particular H200 processors were used for weapons development or military operations.
Finally, how widespread is the practice?
Reuters’ review of 26 records provides evidence of a broader market, but it does not establish the total number of restricted Nvidia processors available to Chinese institutions.
These distinctions are essential.
The story is significant because it reveals potential weaknesses in the enforcement of advanced technology restrictions—not because every unanswered question has already been resolved.
The Bigger Question: Can Washington Control Where Its Most Powerful AI Chips Go?
The latest revelations expose one of the most difficult challenges in the US-China technology rivalry.
Washington can restrict direct exports, require licences and place institutions on trade blacklists.
But controlling access becomes more complicated when advanced processors move through domestic suppliers, third-country intermediaries, server manufacturers and cloud-computing platforms.
For Nvidia, the controversy highlights the political risks of doing business in a market where commercial demand and national-security restrictions increasingly collide.
For China, the findings demonstrate the continuing demand for high-performance AI computing resources, even as the country invests in domestic semiconductor alternatives.
And for the United States, the central issue is whether existing controls can keep sensitive technology away from organisations that Washington considers security risks.
The procurement records have raised serious questions, but they have not established the complete origin of the processors or proven that Nvidia directly supplied the university.
Two H200 chips may appear to be a small purchase in the global AI industry. But if a blacklisted, defence-linked university can obtain restricted computing power through domestic suppliers, how many other institutions could be accessing technology Washington believes it has already blocked?
