Jensen Huang Declines Senate Hearing as Nvidia Counts Its $50B China Gap

Jensen Huang declined a Senate Banking Committee invitation on June 8, 2026. This piece tracks how his public position on US chip export controls evolved from the 2022 A100 ban through the $4.5B H20 charge, the July 2025 policy reversal, and today's $50B China revenue gap in Nvidia's own guidance.

6 min read
Jensen Huang, Nvidia CEO, chip export policy evolution, 2026
Jensen Huang, Founder, President and CEO of Nvidia, at a meeting with EU officials in 2024.· Photo by Peter Dasilva, via Wikimedia Commons (CC BY 4.0)

Jensen Huang declined a Senate Banking Committee invitation to testify on AI chip sales to China on June 8, 2026, the same week Nvidia's forward guidance showed roughly $50 billion in annual China data-centre revenue excluded from its fiscal 2027 numbers, according to CNBC.

Related startups

How the Position Shifted, Year by Year

When the Biden administration's October 2022 export controls banned Nvidia's A100 and H100 chips from China, Huang's public response was commercially pragmatic: China represented one of the world's largest data-centre markets and export controls, his argument ran, risked steering buyers toward non-US suppliers. Nvidia responded by engineering the H20 processor, a downgraded variant specifically designed to comply with the new performance thresholds while keeping a Chinese revenue line open.

By May 2023, Huang was framing the Chinese market as a "long-term opportunity" and part of what he described as a $200 billion addressable market in data centres and CPUs. He joined a US business delegation to Beijing that month, meeting with government and industry officials, and returned with no public criticism of US policy. The posture was calculated: bullish on the market, silent on the restrictions themselves.

That silence became more difficult in April 2025, when the Trump administration required an export licence for the H20 itself. Nvidia took a $4.5 billion excess-inventory charge in the first affected quarter, with Huang later describing the total revenue impact as more than $10 billion. The response was not a press release; it was a lobbying campaign. Huang personally visited the White House and returned to Beijing. He also announced a $500 billion commitment to manufacture AI servers in the United States, timed to the administration's reshoring agenda.

Nvidia annual revenue FY2023 to FY2025 bar chart
Nvidia annual revenue, FY2023-FY2025 ($B). Sources: Nvidia FY2024 10-K and FY2025 earnings via SEC filings.

The H20 Reversal and Its Price

Three months after imposing the H20 licence requirement, the Trump administration reversed course. In July 2025, Nvidia was permitted to resume H20 shipments to China under a revised arrangement, reported by Built In, in which Nvidia and AMD would direct 15 percent of proceeds to the US Treasury. Nvidia's stock rallied on the announcement, and the company projected up to $5 billion in H20 revenue in the following quarter.

Speaking to Fortune in August 2025, Huang described China as a "$50 billion opportunity" in AI infrastructure. That figure captures both the commercial ambition and the structural exposure: China's data-centre AI spending is large enough that any restriction affects Nvidia's consolidated results materially. Nvidia's Q2 fiscal 2027 guidance of approximately $78 billion explicitly excludes China data-centre compute revenue, according to Nasdaq's earnings coverage, meaning the $50 billion annual China run-rate remains structurally off the guided numbers under current policy.

For context on the Chinese AI ecosystem's response, Liang Wenfeng's DeepSeek has demonstrated that Chinese developers can train competitive frontier models with fewer of Nvidia's top-tier chips, a development that complicates the policy case that restricting chip access preserves a durable US lead.

Nvidia sovereign AI revenue FY2025 vs FY2026 bar chart
Nvidia sovereign AI revenue: FY2025 vs FY2026 ($B). Source: Nvidia Q1 FY27 earnings call via Nasdaq, May 2026.

Sovereign AI as the Narrative Replacement

The clearest strategic shift in Huang's public communications since 2022 is the elevation of sovereign AI as a counterweight to China-related uncertainty. Sovereign AI refers to revenue from national governments building domestic AI infrastructure, including computing clusters, model-hosting capacity, and national data centres. In Nvidia's fiscal 2026, sovereign AI revenue crossed $30 billion, up more than three times year-on-year, and accounted for approximately 14 percent of total Nvidia revenue.

Huang's language in earnings calls has tracked this reorientation closely. In 2022 and 2023, his public comments on geopolitics centred on the China opportunity and the competitive risk of driving buyers toward domestic Chinese alternatives. By 2025 and 2026, the forward narrative shifted to agentic AI and sovereign-infrastructure demand. "The agentic AI inflection point has arrived," he stated in the Q1 FY27 earnings release. Countries in Europe, the Middle East, and Southeast Asia building national AI clouds have, in his framing, replaced China as the primary incremental-growth story.

That framing faces its own competitive pressures. Satya Nadella's Microsoft, now reporting $37 billion in annualised AI revenue, is competing for many of the same sovereign-infrastructure contracts. And frontier AI labs including Ilya Sutskever's SSI are building custom-silicon roadmaps that could, over a multi-year horizon, reduce the fraction of AI compute running on Nvidia hardware. Huang's $81.6 billion Q1 FY27 result, up 85 percent year-on-year, suggests sovereign AI is absorbing demand for now; whether it structurally replaces the China market is the question his June 8 posture declined to answer publicly.

Nvidia Q2 FY27 guided revenue versus estimated China quarterly gap doughnut chart
Nvidia Q2 FY27: guided revenue of $78B (ex-China) vs estimated quarterly China run-rate of ~$12.5B. Sources: Nvidia Q1 FY27 earnings (Nasdaq); Fortune, Aug 2025.

What It Means

Huang's June 8 decision to decline Senate testimony is consistent with Nvidia's four-year posture: engage regulators privately, maintain public neutrality on specific policy outcomes, and redirect the earnings narrative toward accessible markets. His offer to host senators at Nvidia's Santa Clara headquarters extends that pattern. The structural question that remains open is whether sovereign AI and the rest-of-world data-centre buildout can fully absorb the China revenue that current export policy keeps off Nvidia's guided numbers. At an estimated $50 billion annually, that gap is not a rounding error; it is a material fraction of what Nvidia's fiscal 2027 run-rate could otherwise show.

Sources

Editorial standards: every claim is sourced. Tips: editor@startuphub.ai

© 2026 StartupHub.ai. All rights reserved. Do not enter, scrape, copy, reproduce, or republish this article in whole or in part. Use as input to AI training, fine-tuning, retrieval-augmented generation, or any machine-learning system is prohibited without written license. Substantially-similar derivative works will be pursued to the fullest extent of applicable copyright, database, and computer-misuse laws. See our terms.