America’s New Weapon Against China’s AI Ambitions: A Tracking Chip in Every GPU

The House Foreign Affairs Committee voted on March 26, 2026, to advance the Chip Security Act (H.R. 3447) — a bipartisan measure that would require manufacturers of advanced AI chips to embed security mechanisms capable of verifying the location of exported chips and detecting potential diversion to unauthorized countries. The committee vote moves the bill to the full House floor for consideration, where it still faces significant legislative hurdles before it could become law.

What the Bill Actually Does

The legislation, introduced in May 2025 by Representatives Bill Huizenga (R-MI) and Bill Foster (D-IL), directs the Secretary of Commerce to establish standards for “chip security mechanisms” — defined broadly to include software, firmware, hardware, or physical security features. Under the bill, chipmakers would be required to verify the ownership and physical location of covered products after export, maintain records of location and end-users, and report credible information about potential diversion of restricted technology.

The bill does not apply to all graphics cards sold commercially worldwide. It targets specifically the high-end AI accelerators classified under U.S. export control classifications — chips like Nvidia’s H100 and H200 that are already subject to licensing requirements when sold to countries of concern.

The Smuggling Scandal That Accelerated the Vote

The committee’s timing was deliberate. On the same day as the markup, the Department of Justice unsealed an indictment in Manhattan federal court charging Yih-Shyan “Wally” Liaw — co-founder of Super Micro Computer — along with Ruei-Tsang “Steven” Chang and Ting-Wei “Willy” Sun with conspiring to illegally ship AI servers containing Nvidia chips to Chinese customers between 2024 and 2025. Prosecutors allege the trio routed U.S.-made servers through Taiwan to Southeast Asia, where they were repacked in unmarked boxes before being forwarded to China, circumventing export controls. The alleged value of the diverted shipments: $2.5 billion.

That case followed a series of similar federal actions. In November 2025, four individuals were accused of running a Florida front company that received $4 million from Chinese firms to purchase and export Nvidia chips. A month later, a separate indictment was unsealed. And just days before the markup, Stanley Yi Zheng, Matthew Kelly, and Tommy Shad English were charged with attempting to procure millions of dollars in restricted chips from a California hardware company; Zheng was arrested March 22, with Kelly and English surrendering three days later.

Nvidia’s Actual Position

Contrary to the viral claim that Jensen Huang said the act would “kill” the industry, no verified statement matching that characterization has emerged. Nvidia’s public posture on this specific bill has been one of cautious opposition rather than dramatic confrontation. An Nvidia spokesperson told reporters that “strict compliance is a top priority,” while the company has argued separately that unlawful diversion of its chips is harmful to its business as well.

Huang’s more fiery rhetoric has been directed at other legislation. In December 2025, he lobbied successfully against the GAIN AI Act — which would have required U.S. chipmakers to prioritize domestic buyers before selling abroad — calling it “even more detrimental to the United States than the AI Diffusion Act.” That bill was dropped from the annual defense legislation after Huang met directly with President Trump and key lawmakers. The “kill the industry” framing circulating online appears to conflate statements made about different bills.

On the Chip Security Act specifically, Nvidia has been more measured. The company has developed its own software-based tracking solution for its Blackwell-generation AI GPUs that it argues can provide operators a rough indication of chip location — suggesting it sees some form of location verification as inevitable, even if it prefers a software approach over legislated hardware mandates.

Senators Escalate Pressure on Export Licenses

Separately, on the same day as the committee vote, Senators Jim Banks (R-IN) and Elizabeth Warren (D-MA) sent a letter to Commerce Secretary Howard Lutnick urging an immediate review — and potential suspension — of all active export licenses covering advanced Nvidia AI chips and server systems destined for China and intermediary countries including Malaysia, Thailand, Vietnam, and Singapore. The senators disputed Huang’s 2025 claims that GPU diversion was not occurring, calling those statements “contradicted by reporting available at the time” and potentially misleading to U.S. officials who relied on them when approving export licenses.

The pressure is particularly acute because the Trump administration reversed a ban on H200 chip exports to China in December 2025, allowing sales under individual licenses. Huang confirmed in March that Nvidia had received orders and restarted production of H200s for the Chinese market. That narrow window is now under scrutiny from both Congress and Reuters reporting that Chinese universities — including some with military affiliations — acquired Super Micro systems with Nvidia A100 chips between 2025 and early 2026 despite existing controls.

What Comes Next

The Chip Security Act now proceeds to a vote by the full House of Representatives. From there it would need to pass the Senate — where a companion bill (S. 1705) has bipartisan backing from Cotton and Warren — and receive a presidential signature before becoming law. Legislative analysts at GovTrack estimate the bill has roughly a 32% chance of advancing out of committee and a 12% chance of full enactment, reflecting the historically low passage rate of most introduced legislation.

The broader debate it has crystallized — how tightly Washington can police its most powerful technology without eroding the commercial dominance that makes that technology worth protecting — is likely to define U.S. semiconductor policy for years to come.