US committee targets Nvidia over Chinese campus ties and Huawei affiliate presence

A US congressional committee has criticised Nvidia for operating on a shared campus in China with a Huawei affiliated company that is under US restrictions. Members say China has been “in Nvidia’s backyard” for a decade. The case is another sign that US China tech policy is moving beyond product lists and export rule text. Lawmakers are now looking at where US chip companies put people and buildings, who they sit next to, and how that might increase the risk of knowledge or talent leaking out.

What the committee is actually worried about

The core issue is location and proximity. Nvidia has been using a campus in China that also hosts a Huawei affiliate that has faced US export controls for years. The committee’s concern is that this raises the risk profile around any work Nvidia does on site, even if that work is formally separated from restricted products.

The allegation is not that Nvidia is handing over GPU blueprints in a conference room. It is softer than that. Lawmakers are worried about the surface area that comes with co location. That includes who employees talk to, how talent moves between companies, and what can be inferred from working in the same environment, even with legal boundaries in place.

In plain terms, the committee sees a gap between US policy toward Huawei on paper and the reality of a US AI hardware leader sharing a physical site with a Huawei affiliate.

How this fits Nvidia’s China balancing act

Nvidia has been walking a fine line in China since the first AI related export controls. Each time the US tightened performance thresholds for export to China, Nvidia responded with new “compliant” GPU variants that sit just under the limits. That strategy kept Chinese data center revenue alive without openly breaking the rules.

Up to now, most Washington criticism has focused on that product dance. The question was whether these China specific SKUs respected the spirit of the controls as well as the letter. The shared campus issue adds a new dimension. It is about the physical footprint that surrounds those products, not just the chips themselves.

From Nvidia’s perspective, running facilities in China is still part of serving local customers. China is a major market for gaming, professional graphics and AI accelerators even after restrictions. Local teams help with support, optimisation and integration. The committee is effectively saying that some site choices cross a line in terms of national security expectations.

Why a shared campus matters in practice

To understand why this triggers such a strong response, it helps to think in terms of information gradients, not movie style IP theft.

Talent and hiring flows – When two advanced tech companies occupy the same campus, staff inevitably build social and professional ties. Over time that makes it easier for employees to switch sides. Lawmakers worry that engineers who understand GPU and interconnect design can move into a Huawei affiliate environment that US policy is trying to isolate.
Shared local ecosystem – Campuses often share facilities, contractors and service providers. That might include network support, security, maintenance, or local IT vendors. Even with strict controls, more shared infrastructure means more potential touch points where operational knowledge can leak.
Signal and perception – In a politicised environment, optics count. To some members of Congress, the decision to share a campus with a Huawei affiliate suggests that Nvidia has not fully internalised how much separation Washington expects between blacklisted Chinese firms and US chip champions.

None of these points automatically imply a clear legal breach. They are about risk tolerance. The committee is effectively arguing that co location with a restricted entity is now an unacceptable level of risk for a company in Nvidia’s position.

What pressure Nvidia could face next

The committee does not write export control rules, but it does shape the political climate around them. The likely outcomes are political and regulatory pressure, not immediate sanctions.

Pressure to move or restructure the site – The simplest lever is public and private pressure to shut down or relocate operations on that campus, or to split facilities so that Nvidia is no longer sharing space with a Huawei affiliate.
Export rules that consider co location – Lawmakers could push the Commerce Department to treat physical proximity to restricted entities as a specific risk factor when evaluating licences or future rules.
More disclosure on global footprints – Nvidia and other chip firms may be pushed to give more detail on where their design, R and D and support centres are located, and who else operates on the same campuses.

All three paths would raise the cost of running a large Chinese footprint. Moving or duplicating sites is expensive. Extra reporting and compliance work pulls management attention away from engineering and product delivery.

Implications for other US chip companies

Nvidia is unlikely to be the only company in this position. Many US semiconductor vendors have engineering, support or joint lab facilities in Chinese tech parks that also host local champions. The committee’s language reads like a warning shot to the entire sector.

We can expect a few shifts.

More conservative site selection – US companies will be more cautious about joining campuses where restricted or sensitive Chinese firms are present. In some cases they will decide that the risk to future US licences or contracts outweighs the efficiency of co location.
Stronger internal security reviews – Boards and security teams will ask harder questions about who shares facilities, how staff move between companies and how third party contractors interact with on site systems.
Separation of sales and core R and D – Some firms may keep customer support and sales presence in China but pull more core design and research activity into jurisdictions that are viewed as safer by US regulators.

This does not stop US chip companies from being global, but it does narrow the options for how that global footprint can be structured without attracting political risk.

Reading the “backyard” language

The committee’s line that China has been “in Nvidia’s backyard” for a decade is revealing. It shows how physical proximity is now being interpreted as strategic exposure. If Nvidia is the key supplier of AI accelerators to US and allied data centers, some lawmakers see any long term co location with a Huawei related firm as a serious blind spot.

From a technical perspective, that framing is blunt. Modern chip design work is distributed across multiple countries. Access to the same campus does not automatically grant access to source code, masks or architecture diagrams. Nvidia’s most sensitive work is still done in the United States and other trusted regions.

But politics will not track those details. To a committee already primed to find examples of US dependence or vulnerability, the image of Nvidia and a Huawei affiliate sharing a campus is enough to justify calls for change.

My take – this is about control of the tech stack, not just one campus

The Nvidia campus story matters, but mostly as part of a wider shift in how Washington thinks about the semiconductor and AI stack.

First – US policymakers now expect large chip vendors to self moderate their China presence. If a site choice or partnership looks risky under the current political lens, they expect companies to avoid it even if there is no explicit rule against it.
Second – The cost of hedging is rising. Nvidia and peers are trying to keep access to Chinese revenue while convincing US regulators that core IP and supply chains are safe. Extra scrutiny on campuses, talent flows and local partners makes that balancing act more complex and more expensive.

If you zoom out, the shared campus is one visible point on a trend line. Export rules already hit Nvidia’s top end products directly. Now lawmakers are probing where the company operates and who it stands next to. Other US chip vendors with big China footprints should expect similar attention. The practical response will be a gradual redraw of global site maps, with more distance placed between US AI and GPU engineering and any entity that sits on a US restriction list.