Nvidia Has No Plan B if China Blockades Taiwan
Nvidia, the world's most valuable chipmaker, reportedly has no contingency plan for a Chinese blockade or invasion of Taiwan, even though nearly all of its advanced AI processors are manufactured and packaged on the island by TSMC. The issue resurfaced on November 9, 2026, when Julian Gewirtz, a former White House China policy official, highlighted passages from Stephen Witt's biography The Thinking Machine on X, arguing that the ongoing Megaspeed scandal puts those details in a new light.
According to Witt's book, CEO Jensen Huang refuses to consider what would happen to his global business if Beijing moved against Taiwan. Witt wrote that he was surprised to discover the company had done no contingency planning at all for that scenario. When he asked Debora Shoquist, Nvidia's executive vice president of operations and the person in charge of the supply chain, about it, she compared the question to asking what she would do if California fell into the ocean, and said Huang had told her "never to think about this question."
Why the Revelation Matters Now
The book was published in 2025, and the passage drew little attention at the time. It is getting a second look because Nvidia now faces the biggest controversy in its history. In late October 2026, The New York Times reported that US intelligence had identified a close relationship between Jensen Huang and Alice Huang, also known as Huang Le, a senior executive at Singapore-based Megaspeed International. The two are not related. On October 31, The Wall Street Journal reported that intelligence services in the United States, Singapore and Taiwan suspected her of working for Beijing. Singapore's Ministry of Home Affairs denied the same day that its officials had expressed such concerns.
On November 5, the Republican chairs of the House Select Committee on China, the House Foreign Affairs Committee and the House Intelligence Committee asked Huang to attend a closed-door briefing on November 18 about his contacts with Alice Huang and about Nvidia's safeguards against foreign espionage. On November 6, the Financial Times reported that a US intelligence memo concluded that China's Ministry of State Security had recruited her to target the Nvidia chief. Beijing has called the accusations groundless.
On Sunday, November 8, Huang admitted in a message to employees that he had been unfaithful to his wife, and Nvidia's board launched an independent investigation into whether its CEO had been the target of a foreign intelligence effort. The board said it had no reason to believe that company secrets or US national security had been compromised, and Huang remains in his role while the review is underway. On November 9, Bloomberg reported that Nvidia's general counsel had intervened in a US Commerce Department probe into suspected chip smuggling by Megaspeed, an inquiry that was later moved to the Justice Department. Nvidia maintains that its dealings with Megaspeed were ordinary business and that it found no evidence of chip diversion.
Megaspeed and the Export Control Question
Megaspeed was created in Singapore in 2023 as an offshore unit of 7Road, a Chinese gaming and cloud computing company. It grew into the largest buyer of Nvidia chips in Southeast Asia, purchasing more than $4.8 billion worth of hardware since June 2024 according to press reports. Authorities in Singapore have been investigating the firm since October 2025 for suspected breaches of local laws, and US officials have examined whether restricted processors reached Chinese customers through its data centers in Malaysia and Indonesia. In December 2025, reporting found that Megaspeed had imported tens of thousands more Nvidia chips than Nvidia had recorded in its facilities.
The scandal has hit Nvidia's market value hard. Its shares have lost about 21% over the past month, and its market capitalization has dropped to just over $4.5 trillion from a peak of more than $5.7 trillion in October 2026.
A Single Point of Failure in Taiwan
For supply chain managers, the more lasting concern is not the personal scandal but what it reveals about risk governance at the company that sits at the center of the global AI buildout. Nvidia designs its chips but does not manufacture them. Its flagship Blackwell and upcoming Vera Rubin processors are produced by TSMC, mostly at its advanced fabs in southern Taiwan, including the Tainan campus where most of its 4-nanometer and 3-nanometer output is concentrated.
TSMC's Fab 21 in Arizona began producing Blackwell wafers in October 2025 and has since moved to volume production. However, every one of those wafers still has to be shipped back to Taiwan for advanced packaging, because all of TSMC's CoWoS packaging capacity is located on the island. Amkor's packaging plant in Arizona is not expected to start production before early 2028. In practice, no high-end Nvidia data center GPU can be completed today without passing through Taiwan.
Nvidia's exposure has also grown rather than shrunk. In June 2026, Huang said the company was spending roughly $100 billion a year in Taiwan, up from $10 billion to $15 billion a few years earlier, with a target of $150 billion. The company also opened a new Taiwan base in 2026 to support those investments.
Rising Military Pressure Around the Island
The lack of a plan stands out because the risk is no longer theoretical. In late December 2025, the Chinese military staged its largest drills around Taiwan in years, rehearsing a blockade of the deep-water ports of Keelung and Kaohsiung with about 10 hours of live fire. Taiwan's defense ministry warned in its 2026 report on Chinese military power that recent exercises increasingly focus on cutting the island off from outside air and sea routes. In August 2026, Taiwan's annual Han Kuang war games included its first joint navy and coast guard counter-blockade drill.
Huang's public statements on the subject have been brief. In 2023, he said Nvidia felt perfectly safe relying on Taiwan. In September 2024, he told a Goldman Sachs conference that supply would continue if TSMC were compromised, because Nvidia holds enough intellectual property to move production to other foundries, while admitting that the resulting chips would not be as good. Industry analysts note that no other foundry currently matches TSMC's leading-edge process and packaging capacity at the volumes Nvidia needs.
What It Means for AI Chip Buyers
A blockade of Taiwan would cut off the supply of advanced AI accelerators for cloud providers, AI labs, server makers and governments worldwide, with knock-on effects on memory, servers and data center projects. The House briefing on November 18 and the board's investigation are expected to focus on espionage and export controls, but investors and customers are increasingly asking whether Nvidia has a credible business continuity plan for its most concentrated supply risk.
Until advanced packaging capacity outside Taiwan comes online and Nvidia qualifies alternative manufacturing routes, the global AI hardware supply chain will continue to depend on a single island that Beijing openly trains to isolate.
💡 Alternative Solution
Formal business continuity planning for a Taiwan blockade scenario, faster ramp-up of Blackwell and Rubin wafer output at TSMC Arizona, CoWoS-class advanced packaging in the US through Amkor and TSMC Arizona, qualification of second-source foundries such as Samsung and Intel for some product lines, use of Intel EMIB and Foveros packaging, strategic buffer stocks of finished GPUs outside Taiwan, diversification of HBM and substrate sourcing toward South Korea, Japan and the US, stronger end-use checks on Southeast Asian distributors