A US investigation into Megaspeed, a Singapore-based AI cloud services company, has revealed increasingly complex commercial risks behind the globalization of AI computing power. According to The New York Times' Oct 29 report, the US government is investigating Megaspeed over suspected evasion of AI chip export controls and has launched an intelligence investigation into its former executive Alice Huang (also known as Huang Le). US authorities suspect she may have used her business ties with Nvidia CEO Jensen Huang to obtain commercial information and pass intelligence to China.

Citing records from trade data platform ImportGenius, the report said Megaspeed's affiliated company in Malaysia purchased at least US$4.8 billion worth of advanced AI equipment containing Nvidia chips and related technology between June 2024 and February 2026. The scale of the purchases has drawn attention to the ultimate destination of the equipment, its actual use, and the compliance of the cross-border computing supply chain.

It should be noted that US$4.8 billion is the purchase value of the equipment, not a verified amount of illegal transshipment. So far, public information is insufficient to prove that Megaspeed has illegally transshipped the equipment to China, nor can the intelligence allegations against Alice Huang be treated as judicially established facts.

The case reflects a key challenge facing the AI infrastructure industry: as high-performance GPUs, data centers and cloud computing services globalize at an accelerating pace, companies may face export controls and national security reviews due to end users, equipment use or cross-border business relationships, even when their procurement, deployment and sales activities each occur in different countries.

Behind the US$4.8 Billion Equipment Purchases, Cross-Border Computing Supply Chains Face Scrutiny

Megaspeed is a Singapore-headquartered AI cloud services and data center company whose business involves AI computing infrastructure. Such companies participate in the global AI infrastructure market by procuring high-performance GPUs, building data centers and providing computing services to customers.

However, amid US export controls on advanced AI chips, GPU commercial transactions are no longer ordinary deals between suppliers, cloud providers and customers. Who ultimately uses the equipment, where it is deployed, and whether computing power is resold to restricted entities can all affect the compliance of a transaction.

According to The New York Times, the US government has been investigating Megaspeed since late 2024, suspecting the company may have helped Chinese enterprises obtain computing resources from export-controlled AI chips by deploying data centers in Southeast Asia and reselling related equipment.

The report said US Commerce Department officials visited Megaspeed's data center in Malaysia and found some Nvidia equipment still in packaging, raising questions about the actual deployment and final use of the equipment.

In 2025, Singapore authorities also questioned Alice Huang over related matters before releasing her. In 2026, the US Commerce Department's investigation into Megaspeed was reportedly handed over to the Department of Justice.

These leads do not mean the equipment has been confirmed to have been illegally transshipped, but they show regulators' scrutiny of the AI computing supply chain extending from chip procurement to equipment deployment, data center operations and end-user identification.

For AI cloud providers, compliance in equipment procurement does not necessarily mean the entire business chain is compliant. Even when companies purchase GPUs from legitimate channels, they still need to pay attention to where equipment is deployed, customer identities, actual usage scenarios, and whether related transactions comply with applicable export control regulations.

From Chip Sales to Intelligence Investigations, Business Relationships Can Also Become a Source of Risk

Another lead in the case involves business dealings between Megaspeed's former executive Alice Huang and Nvidia CEO Jensen Huang.

According to The New York Times, citing anonymous sources, US intelligence agencies monitored Alice Huang for about a year, investigating her ties to China. US officials feared she might use her contacts with Jensen Huang to obtain Nvidia's business plans, as well as information about Jensen Huang's communications with the Trump administration on AI chip export policy toward China.

The report said Trump received the intelligence briefing about a year ago but took no action. Sources said the decision may have involved considerations of Nvidia's commercial interests, US stock markets and US-China trade relations.

The above content comes mainly from anonymous sources, and the intelligence allegations have yet to be confirmed by complete public evidence.

Nvidia denies any anomaly in the relationship between Jensen Huang and Alice Huang. A company spokesperson said Jensen Huang communicates with thousands of customer executives every year, and his contact with her was normal business interaction. Nvidia also said regional cloud services customers receive routine commercial information rather than sensitive trade secrets or national security-related information.

The controversy also shows that in the AI industry, commercial risk and national security issues can intersect. Large chip companies, cloud providers and data center operators usually need to maintain extensive ties with customers, investors, government agencies and industry partners. But when business involves controlled technology, cross-border computing supply or sensitive markets, some business relationships may also become the subject of regulatory scrutiny.

This does not mean normal customer relationships themselves constitute risk, but rather that companies need to draw clearer boundaries between commercial and sensitive information, and establish corresponding information access controls, customer review and internal compliance mechanisms.

As AI Infrastructure Globalizes, Companies Face Three Types of Commercial Risk

The issues raised by the Megaspeed case have industry-wide significance beyond any single company. For AI chip suppliers, cloud providers, data center operators and AI application companies expanding globally, the following three types of risk deserve particular attention.

First, export control risk may extend from hardware transactions to computing services.

The cross-border flow of AI chips is only part of the regulatory chain. Equipment deployment, resale, leasing, computing services and changes in end users can all involve different legal obligations.

For companies, the risk lies not only in whether they directly sell GPUs to restricted customers, but also in whether transaction structures comply with applicable regulations and whether they can continuously track equipment end use and customer changes. Once suspected of wrongdoing, companies may face investigations, blocked transactions, equipment unable to be deployed or customer attrition.

Second, Southeast Asian data centers may face higher compliance and operating costs.

Southeast Asian markets such as Singapore and Malaysia hold an important position in the global AI infrastructure landscape. Companies can expand by building data centers, procuring equipment and providing cloud services across different countries, but cross-border presence also adds complexity in legal applicability, customer identification and supply chain management.

If regulators question the ultimate destination of equipment, companies may need to invest additional resources to prove the compliance of procurement, transportation, deployment and service records. For capital-intensive data center businesses, idle equipment, project delays or blocked financing can all affect returns on investment.

Third, geopolitical risk may affect company valuations and business partnerships.

Advanced AI chips carry both commercial and strategic value, making related companies more susceptible to export controls, national security reviews and shifts in diplomatic relations.

Once a company or its executives come under major investigation, even if no violation is ultimately established, they may face customers re-evaluating partnerships, suppliers raising due-diligence requirements, and investors demanding higher risk compensation.

For companies that need to continuously invest huge capital in GPU procurement and data center construction, this uncertainty is especially important: the viability of a business model depends not only on computing demand, equipment utilization and customer willingness to pay, but also on whether the company can keep operating across different jurisdictions.

AI Infrastructure Investment Logic Needs to Incorporate Compliance Capability

In the past, discussions of AI infrastructure investment revolved mainly around GPU supply, data center construction, power costs, computing demand and customer growth. But as the strategic value of advanced computing technology rises, regulatory risk should also become an important dimension in evaluating a company's competitiveness.

For investors, in addition to revenue growth, capital expenditure and equipment utilization, it is necessary to further examine a company's customer structure, end-user identification mechanisms, cross-border equipment management capabilities, and its ability to adjust operations when export controls change.

For company managers, compliance should not be just a review procedure after transactions are completed, but should be embedded in customer onboarding, contract design, equipment deployment, computing delivery and ongoing operations. Particularly in cross-border cloud computing, companies need to clarify which laws apply in different regions, keep complete transaction and equipment records, and build internal mechanisms that can respond promptly to regulatory changes.

Key facts in the Megaspeed case still await further confirmation. Nvidia denies any anomaly in the relevant business dealings, and Megaspeed has said it is willing to cooperate with the US government investigation. At this stage, public reports are insufficient to conclude that the company has engaged in illegal chip transshipment, nor can Alice Huang be said to have been confirmed by judicial authorities to have engaged in intelligence activities.

But from an industry perspective, the case has highlighted a trend worth noting: the globalization of AI computing power is not only a competition of capital, technology and infrastructure, but also a competition of supply-chain transparency, cross-border compliance and risk management capability. For AI infrastructure companies, being able to deliver computing power to target customers sustainably and compliantly may be as important as securing chips, building data centers and expanding market share.

References:
[1] TechCrunch, US investigated Chinese spy plot aimed at Nvidia CEO;

[2] The New York Times, Nvidia Chief Seen as Target of Chinese Spy Plot;

[3] The Straits Times, Nvidia CEO Jensen Huang says ties to company under spying probe were ‘normal’;

[4] The Guardian, Nvidia CEO reportedly targeted by an alleged Chinese spy