Nvidia’s AI Dominance Faces a Test of Trust and Money

Espionage allegations involving a Singapore-based customer, restrictions on chip sales to China and questions over circular financing are exposing new vulnerabilities in the world’s leading AI chipmaker.

nVIDIA

Opinions expressed by Entrepreneur contributors are their own.

You're reading Entrepreneur Asia Pacific, an international franchise of Entrepreneur Media.

For a company that has made billions selling the technology behind the artificial intelligence boom, Nvidia is discovering that dominance comes with risks extending far beyond the products it makes.

The American chipmaker, whose processors run everything from OpenAI’s ChatGPT to advanced AI research, now faces pressure on three fronts.

A board investigation is examining reports that Chinese intelligence targeted its chief executive, Jensen Huang. Its efforts to keep restricted chips out of China are under growing scrutiny. Investors and lenders are also questioning how Nvidia helps fund the businesses that buy its technology.

Investors reacted quickly. Nvidia shares fell 4.3% on November 9, wiping about $190 Bn off its market value on a day the broader U.S. market rallied, after the board announced an independent investigation into the reports.

Nvidia and Huang were also sued on November 9 in a proposed shareholder class action alleging that they misled investors by concealing Huang’s reported relationship with Alice Huang and the associated national-security concerns. The lawsuit, filed in a California federal court, seeks unspecified damages.

In its November 8 statement, the board said a committee of independent directors, with help from outside counsel, would lead the inquiry. It said it had no reason to believe national security or Nvidia’s confidential information had been compromised and reiterated its confidence in Huang.

The investigation follows reports that Huang had a close relationship with Alice Huang, a former executive of Megaspeed International, a Singapore-based cloud computing provider and Nvidia customer. U.S. intelligence agencies reportedly suspected her of gathering information for China.

In a message to employees, Huang admitted he had been unfaithful to his wife, without naming Alice Huang. China has dismissed the spying allegations as unsubstantiated.

The affair has also reached Washington, D.C. Three senior Republican lawmakers have asked Huang for a closed-door congressional briefing on November 18 to explain his dealings with Alice Huang and Nvidia’s safeguards against foreign espionage. The date falls a day after Nvidia is scheduled to report its third-quarter results.

The matter is sensitive because Huang is no ordinary chief executive. He has led Nvidia since its founding in 1993 and has become one of the most influential voices in the US-China technology contest.

The immediate question is whether Nvidia’s security procedures were adequate. The bigger commercial worry is that heightened political scrutiny could make it harder for the company to sell into international markets.

Singapore and the Chip Control Problem

The investigation has also put ASEAN in the spotlight. Megaspeed has separately drawn the attention of U.S. authorities over suspicions that advanced American AI chips may have been diverted to China in breach of export restrictions. The company denies wrongdoing.

In a statement dated December 23, 2025, Megaspeed said inspections and reviews by U.S. and Malaysian authorities, as well as Nvidia, had found no evidence of illegal chip transfers or other violations.

The Singapore Police Force has confirmed that investigations into Megaspeed are ongoing, while the Ministry of Home Affairs has explicitly denied that Singapore authorities investigated Alice Huang over her alleged intelligence links.

Singapore police have also said the city-state does not implement export controls imposed unilaterally by other countries, but acts firmly against companies and individuals that break its own laws. They added that companies operating in Singapore are expected to take other countries’ regulations, including export controls, into account in their international business activities.

None of the publicly disclosed investigations has established either an espionage operation involving Nvidia or illegal chip transfers by Megaspeed. The episode does expose a practical problem for Nvidia. Selling advanced chips is the easy part; making sure they stay with authorised users after passing through cloud providers and data centres in several countries is far harder.

Bloomberg reported on November 6 that Nvidia was tightening controls on its Southeast Asian cloud partners. Companies seeking official Nvidia Cloud Partner status must now enrol in a system that tracks where their AI processors are located, according to people familiar with the arrangement.

The requirement is reportedly tied to Nvidia’s partner programme and is not a blanket rule for every customer. Even so, it shows the pressure on the company to track its products more closely after they leave the factory.

For ASEAN data centre operators, tighter tracking and compliance could add to operating costs and complicate dealings with overseas customers. For Nvidia, weak controls risk inviting more regulatory intervention.

China is Already Costing Nvidia Money

The pressure goes beyond espionage claims and chip diversion. Nvidia’s most advanced processors sit at the centre of the technology rivalry between Washington and Beijing. The US restricts exports of certain high-performance chips to China to limit their use in advanced computing, while Beijing promotes domestic alternatives to reduce its dependence on American technology. Nvidia is caught in the middle.

On November 5, Reuters reported that about 600,000 Nvidia H200 AI chips were being held in Hong Kong, awaiting Beijing’s approval for large-scale imports into mainland China. The stockpile shows how uncertain the market has become: customers want the computing power, but governments decide who gets it.

The cost already shows up in Nvidia’s accounts. The company disclosed a $400 Mn charge in the first half of its fiscal year 2027 for excess H200 inventory and purchase obligations as demand for the H200 diminished. That followed a far larger $4.5 Bn charge on its H20 processors in 2025, when U.S. export licensing rules disrupted sales. Nvidia’s revenue guidance of $108 Bn for the quarter ended in October assumes no data centre compute revenue from China.

Nvidia has also acknowledged that being effectively shut out of China’s data centre computing market has helped its competitors build larger developer and customer ecosystems to challenge it worldwide.

The impact could last longer than a lost quarter of sales. Nvidia gains not only when customers buy its processors but also when developers build applications around its software. Once businesses build their systems on competing technology, winning them back gets harder.

Is Nvidia Helping Finance Its Own Demand?

Export restrictions threaten access to some markets. A separate concern is emerging over how the AI boom itself is being paid for.

Nvidia is best known for designing graphics processing units, or GPUs, the chips that handle the vast volume of calculations needed to train and run AI systems.

The processors are expensive, and they need data centres, electricity, cooling and other infrastructure to work. Many AI developers and specialist cloud providers cannot build such facilities without outside funding, and Nvidia is increasingly the one providing it.

In January 2026, it invested $2 Bn in CoreWeave, a U.S. cloud computing company that runs infrastructure built on Nvidia technology, and the two expanded their cooperation on hardware, software and data centre development. CoreWeave said the money would not be used to buy Nvidia processors, so the deal does not show Nvidia directly financing purchases of its own chips.

The relationship goes deeper. Under another agreement in September 2025, with an initial value of $6.3 Bn, Nvidia committed to buying CoreWeave computing capacity left unsold to other customers, subject to contractual conditions.

Arrangements like this draw scrutiny because Nvidia is playing roles that are usually kept apart. It supplies the computing equipment, invests in the businesses that operate it and, in some cases, agrees to buy their computing services if they cannot find enough customers.

Critics call this pattern circular financing. The concern is simple: when a supplier helps finance its customers, or guarantees demand for their services, those customers can keep buying more of the supplier’s products. Nvidia’s chief financial officer, Colette Kress, addressed the label on the company’s August earnings call: “We recognise the scale of this support, and we know some will call this circular financing. We see it differently.”

None of this means the sales are fictitious or the deals improper. But it raises the question of how much demand comes from customers paying their own way and how much rests on financial support from within the industry.

Nvidia says in its quarterly filing that its partnerships with AI clouds widen access to its infrastructure for start-ups, model builders, enterprises, research organisations and sovereign customers, and that it is helping select customers secure land, power and data centre capacity. The scale of its commitments shows how tightly its own finances are now tied to the wider AI economy.

Nvidia’s Financial Exposure at a Glance

Selected investments, commitments and guarantees disclosed in Nvidia’s quarterly filing for the period ended July 26, 2026, including subsequent arrangements announced in August.

Financial ArrangementAmountWhat It Means
Equity investments$99 BnNvidia’s disclosed investments in other companies, exposing it to changes in their value
Future equity investment commitments$25 BnCommitments to make additional investments over time
AI cloud agreements$36 BnCommitments to obtain computing capacity from selected cloud partners that buy Nvidia infrastructure
OpenAI-related infrastructure guaranteesUp to $105 BnConditional support for specified lease and power obligations at an Ohio data centre development
Supply and manufacturing capacity commitments$279 BnAgreements to secure components and production capacity for expected future demand

Source: Nvidia Form 10-Q for the quarter ended July 26, 2026, filed Aug 26, 2026.

Note: These amounts represent different types of financial exposure and must not be added together as a single liability or loss. The $105 Bn guarantee is capped, conditional and phased. Some supply agreements can be cancelled, rescheduled or adjusted before firm orders are placed, though changes may add costs.

When Wall Street Starts Asking Questions

Lenders are pushing back too. On October 1, Reuters reported that banks and credit investors wanted stronger guarantees on some proposed AI infrastructure financing. The disagreement centres on how long expensive AI processors keep their commercial value.

Nvidia argues its advanced chips stay productive for many years. Lenders are more cautious, because each new generation of processors can make older hardware less competitive. That matters when chips serve as collateral: if a borrower defaults, lenders need the equipment to hold enough value to recover their money, and the pace of technological change makes that hard to judge.

Nvidia has also disclosed memoranda of understanding with major capital providers aimed at mobilising more than $500 Bn in third-party financing for AI infrastructure over time. These are not completed deals, and Nvidia warns that preliminary agreements may not lead to definitive ones.

The broader risk is that spending on AI infrastructure outpaces the profitable uses businesses find for the capacity. If that happens, cloud providers could struggle to fill data centres, borrowers could struggle to repay, and AI infrastructure investments could lose value. Nvidia could then face weaker chip demand just as some of its investments and contractual exposures come under strain.

That risk is not evidence of an impending financial crisis. Nvidia’s operating performance remains very strong. Revenue for the quarter ended July 26 reached $96.2 Bn, more than double a year earlier, with $89 Bn from its data centre business and a gross margin of 75%.

Those numbers show the business’s scale and profitability. They do not settle whether today’s heavy investment in AI infrastructure will earn enough to pay back the companies financing it.

Can Nvidia Manage the Risks of Its Own Success?

Not every challenge is linked, though. The espionage allegations concern security and corporate governance; export restrictions affect market access and compliance; and circular financing raises questions about whether AI infrastructure spending will last. They converge on one company because Nvidia now sits at almost every stage of the AI industry.

That position brings big advantages. Its processors are everywhere, its software ecosystem is deeply entrenched, and its balance sheet lets it back the expansion of customers and partners. It also leaves the company exposed.

A drawn-out investigation into Huang could raise governance questions, because investors tie Nvidia’s leadership and strategy so closely to its founder. Tighter export controls could restrict sales, and tracking requirements could make doing business abroad more complicated. A slowdown in AI spending would hit from another direction, cutting demand for processors while Nvidia still carries exposure to the companies and projects it has helped finance.

These are not foregone conclusions. The board’s investigation has just begun; Nvidia’s financial commitments are not evidence of improper transactions and demand for its technology remains strong.

What matters now is whether Nvidia can keep expanding the AI market without taking on ever more responsibility for financing, policing and absorbing the risks of that expansion. For years, it could measure success largely by demand for its chips. Its next test is how well it manages the far larger business that has grown up around it.

For a company that has made billions selling the technology behind the artificial intelligence boom, Nvidia is discovering that dominance comes with risks extending far beyond the products it makes.

The American chipmaker, whose processors run everything from OpenAI’s ChatGPT to advanced AI research, now faces pressure on three fronts.

A board investigation is examining reports that Chinese intelligence targeted its chief executive, Jensen Huang. Its efforts to keep restricted chips out of China are under growing scrutiny. Investors and lenders are also questioning how Nvidia helps fund the businesses that buy its technology.

Related Content