The United States government will directly profit from American AI chip exports to China under a groundbreaking new arrangement. According to exclusive reporting by The Financial Times (July 2024), NVIDIA and AMD have agreed to surrender 15% of their revenue from artificial intelligence chip sales to Chinese customers to the U.S. Treasury. This unprecedented revenue-sharing deal, negotiated with the Trump administration, secured export licenses allowing limited sales of advanced AI processors to China – including NVIDIA’s H20 GPU and AMD’s MI308 accelerator – despite sweeping 2023 export controls.

The agreement represents a strategic compromise: U.S. chipmakers regain partial access to China’s massive semiconductor market, while the government creates a new revenue stream tied to restricted technology transfers. NVIDIA confirmed adherence to government rules but didn’t dispute the FT’s report, stating: “We follow the rules the US government sets for our participation in worldwide markets.” AMD declined comment. Analysts estimate this arrangement could generate over $2 billion for the U.S. government in 2025 alone based on NVIDIA’s pre-restriction Chinese revenue of $4.6 billion in a single quarter.

How Will the 15% China Chip Fee Impact US Tech Companies?

The financial burden is significant but calculated. For NVIDIA, which derived 12.5% of its total revenue from China last year, the fee eats into profits from a critical market. However, the alternative – a complete sales ban – proved costlier. NVIDIA previously absorbed a $1 billion quarterly revenue hit from initial restrictions, far exceeding expectations. AMD, facing a sharp operating income decline partly attributed to China constraints, awaits its own export licenses despite CEO Lisa Su’s recent comments about pending approvals.

Industry experts note this creates a complex competitive landscape:

Revenue Trade-off: Companies forfeit 15% of China AI chip sales but regain market share
Pricing Strategy: Costs may be passed to Chinese buyers, making U.S. chips less competitive versus domestic alternatives like Huawei’s Ascend
Government Reliance: Future market access hinges on political negotiations, adding uncertainty

“These licenses are lifelines with strings attached,” remarked Paul Triolo, Associate Partner for China at Albright Stonebridge Group. “The revenue share is effectively a specialized tariff ensuring the US benefits financially from controlled technology exports.”

Where Will the AI Chip Revenue Funds Be Directed?

While the Treasury’s specific allocation remains undisclosed, analysts point to two likely priorities informed by Trump administration policy documents:

Reducing the Trade Deficit: Directly offsetting the U.S.-China goods trade imbalance, which hit $279 billion in 2023 (U.S. Census Bureau)
Boosting Domestic Semiconductor Manufacturing: Subsidizing CHIPS Act initiatives or new fab construction grants

The arrangement coincides with tightened restrictions on advanced High Bandwidth Memory (HBM) chips – critical AI GPU components using U.S. technology – further complicating Chinese firms’ ability to build competitive systems even with approved processors.

This landmark revenue-sharing model transforms how America manages strategic tech exports. By monetizing restricted sales, the U.S. gains economic leverage while allowing chipmakers limited market access – a high-stakes balancing act between national security and industrial interests that could redefine global tech trade dynamics for years to come. Monitor official Treasury disclosures for fund allocation details as this geopolitical tech deal unfolds.

Must Know

Q: Why did NVIDIA and AMD agree to pay 15% to the US government?
A: This revenue share was a condition for receiving special licenses to sell restricted AI chips (NVIDIA H20, AMD MI308) to China. Without it, both faced total bans in this critical market after 2023 export controls.

Q: How much money will the US government make from this deal?
A: Based on NVIDIA’s pre-restriction China revenue, analysts project over $2 billion in 2025. Final amounts depend on actual sales volumes of approved chips.

Q: Can China develop its own AI chips to avoid these fees?
A: Yes, companies like Huawei and SMIC are accelerating domestic AI chip development. However, U.S. restrictions on advanced manufacturing equipment and HBM memory imports create significant technical hurdles.

Q: Will other countries implement similar tech revenue-sharing models?
A: This is the first major instance. Its success may inspire similar “managed access” frameworks globally for sensitive technologies like AI, quantum computing, or biotechnology.

Q: How does this impact Chinese tech companies’ AI capabilities?
A: While they gain access to some advanced U.S. chips, the 15% cost hike and parallel HBM restrictions will slow large-scale AI deployment compared to unfettered access scenarios.

Q: Have these export licenses officially been granted?
A: NVIDIA confirmed receiving licenses per Reuters (July 2024). AMD’s status remains pending per CEO Lisa Su’s recent statements.