Singapore pressed to adopt AI-led checks after Nvidia case

Authorities are under pressure to boost spot checks on high-risk tech shipments.

Singapore may need to move beyond self-declared export controls and adopt artificial intelligence (AI)-driven monitoring after a $3.2b (US$2.5b) scheme involving restricted Nvidia Corp. chips exposed weaknesses in shipment verification.

“Self-declaration is not broken, but it is no longer sufficient in high-stakes sectors without AI-enabled verification and continuous monitoring,” Lee Kuan Yew School of Public Policy Practice Professor Vu Minh Khuong told Singapore Business Review in an emailed reply to questions. “The lesson is clear: Singapore must move from assumed trust to verified trust.”

The case stems from a US Department of Justice indictment on 19 March against Super Micro Computer, Inc. co-founder Yih-Shyan Liaw and two others accused of using an unnamed Southeast Asian company to reroute servers powered by Nvidia chips to China.

Prosecutors alleged that hair dryers were used to remove labels from genuine servers and attach them to dummy units that remained in the region, whilst the actual hardware was shipped to China in plain boxes.

Singapore authorities later charged Aperia International Pte. Ltd. Chief Financial Officer Jenny Lim on of taking part in a criminal conspiracy involving false representations, making her the fourth executive charged in connection with the case.

The US has restricted exports of Nvidia’s most advanced chips to China since 2022 to limit Beijing’s access to advanced computing technology.

Dr Rohit Bhatnagar, associate professor of operations management, ITOM Division at Nanyang Business School, said Singapore’s “light-touch” controls might have created gaps that allowed restricted chips disguised as AI servers to be re-exported.

“This is a regulatory blind spot that needs to be reviewed and assessed deeply, and stricter checks need to be put in place,” Bhatnagar said in an emailed reply to questions.

Singapore regulates strategic exports through the Strategic Goods (Control) Act, which governs the transfer, transhipment, and brokering of sensitive goods and technologies. The law also contains catch-all provisions covering items linked to weapons of mass destruction.

Bhatnagar said advanced AI servers should be classified as “highly risky,” with stricter certification requirements for assemblers and exporters.

Vu said Singapore should adopt a “verified trust” model built on AI-based monitoring, digital chain-of-custody tracking, and risk-based inspections.

He said predictive systems could identify anomalies in shipment patterns and help authorities target physical spot checks more effectively.

Despite the scale of the alleged scheme, both professors said Singapore’s reputation as a trusted trade hub is unlikely to suffer lasting damage.

“What global firms value most is not the absence of incidents, but predictable, efficient, and credible governance,” Vu said. “This is a strategic opportunity rather than a reputational setback.”