Fall ResetAmazon USFall reset deals: check better picks before checkoutAmazon US: today's deals, useful picks and quick comparisons.Check DealsClean PCRecommendedOne scan can reveal what keeps slowing WindowsLook for cleanup and repair opportunities.Run ScanFall ResetAmazon USWork and home upgrades are worth comparing todayAmazon US: today's deals, useful picks and quick comparisons.See Picks×
Skip to content
Blog

NVIDIA and AMD’s 15% China AI-Chip Revenue Deal, Explained

By TheFinanceBase Team6 min read
Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.

Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.

Short answer: In August 2025, the Trump administration linked U.S. export approvals for NVIDIA’s H20 and AMD’s MI308 AI accelerators to an expectation that the government would receive 15% of revenue from licensed China sales. That was reported and publicly confirmed, but the public record reviewed through August 2026 does not show a generally applicable regulation codifying a 15% payment. NVIDIA calls it a government “expectation,” while AMD describes a 15% revenue transfer in its risk disclosures.

The arrangement was narrow: it concerned specified, licensed products—not every NVIDIA or AMD chip, all China revenue, or a conventional tax. It also did not guarantee that Chinese customers would buy the products or that either company remitted a particular dollar amount.

What happened in August 2025?

The episode followed a sharp change in U.S. export policy. On April 9, 2025, the U.S. government told NVIDIA that H20 exports required a license for China, Hong Kong, Macau and specified D:5-related destinations. NVIDIA subsequently recorded a $4.5 billion charge tied to H20 excess inventory and purchase obligations after demand was restricted. NVIDIA’s April 2025 filing documents the licensing requirement and charge.

Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.

As the administration began allowing some shipments again, reporting in August described a deal under which NVIDIA and AMD would provide 15% of revenue from licensed H20 and MI308 sales to the U.S. government. President Donald Trump publicly confirmed the figure and said he had initially sought 20%. The Associated Press reported the arrangement and the administration’s explanation at the time. AP coverage

#1 Best Overall
waveshare Hailo-8 M.2 AI Accelerator Module, Compatible with Raspberry Pi 5, Supports Linux/Windows Systems, Based On The 26TOPS Hailo-8 AI Processor, Module Only
  • ✅Powered by 26 Tera-Operations Per Second (TOPS) Hailo-8 AI Processor. 2.5W typical power consumption
  • ✅Scalable, enabling simultaneous processing of multi-streams & multi-models
  • ✅Enabling real-time, low latency and high-efficiency AI inferencing on the edge devices
  • ✅Supports TensorFlow, TensorFlow Lite, ONNX, Keras, Pytorch frameworks
  • ✅Supports Linux and Windows. Supports the temperature range of -40°C to 85°C

NVIDIA’s later filing is important because it says the government had not published a regulation codifying the 15% requirement. That makes the most accurate description an unusual revenue-sharing expectation or licensing condition associated with export approvals—not an ordinary, generally applicable export fee.

Which sales did the 15% concern?

Company Product Scope described in public filings and reporting
NVIDIA H20 Licensed H20 sales to China; the product was designed to comply with earlier performance limits and later became license-controlled.
AMD MI308 Licensed MI308 sales to China; AMD’s annual report describes a 15% government share of revenue generated by those sales.

“Revenue” is not the same as profit, cash actually remitted, shipment value, or total China sales. The public materials reviewed do not establish a universal formula for calculating the share, a payment schedule, or a confirmed remittance by either company.

NVIDIA H20

The H20 is a China-focused accelerator derived from NVIDIA’s Hopper generation and configured to meet earlier U.S. performance restrictions. After the April 2025 license requirement, NVIDIA disclosed the $4.5 billion charge noted above. Later licenses permitted at least some H20 shipments, but Chinese authorities and customers subsequently raised security concerns about the product. The Congressional Research Service summarizes the policy and subsequent concerns.

Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.
Rank #2
MX3 M.2 AI Accelerator
  • High-Performance AI Processing: The MX3 is designed to handle the most demanding AI computer vision workloads, delivering exceptional performance and efficiency.
  • Flexible Integration: The MX3 can be easily integrated into your existing systems via its M.2 M-key form factor and support for Linux operating systems.
  • Energy Efficient: The MX3 is designed to provide high performance while minimizing power consumption.
  • Comprehensive Software Development Kit (SDK): The MX3 is supported by a comprehensive SDK that simplifies development and deployment.
  • Hardware compatability: The MX3 is compatible with the PCI-SIG M.2 M-key 2280 Specification. It can be used with the Raspberry Pi 5 with a M-key 2280 HAT.

AMD MI308

The MI308 is AMD’s China-relevant accelerator derived from the Instinct MI300 family. AMD’s 2025 annual report says the U.S. government would receive 15% of revenue from licensed MI308 sales to China and warns that the arrangement could lead to litigation, higher costs and reduced competitiveness. AMD’s annual report

Is the 15% a tax, tariff, fee or donation?

None of those labels is fully established by the public record.

  • Not clearly an ordinary tax: The arrangement was tied to particular export permissions and products, rather than a generally applicable tax enacted by Congress.
  • Not a conventional tariff: It was associated with export licensing, not a customs duty on imports.
  • Not clearly voluntary: A company’s ability to sell depended on obtaining a U.S. license, which makes a “donation” characterization contestable.
  • Best description: A reported revenue-sharing expectation or licensing condition connected to export approvals.

The Congressional Research Service said Congress could examine the executive branch’s authority to receive proceeds from licensed chip sales. Lawmakers and commentators questioned whether conditioning export permission on a government revenue share fits existing export-control and appropriations authorities. Those are legal and constitutional questions—not a court ruling that the arrangement was invalid. CRS analysis

How much money was involved?

Early estimates illustrate the difference between potential and realized money. One contemporary estimate put possible 2025 H20 China sales at about 1.5 million chips and roughly $23 billion in revenue if restrictions had not intervened. A 15% share of that estimate would be approximately $3.45 billion, but that is a hypothetical calculation based on estimated sales—not money shown as paid to the government. The estimate was reproduced in contemporaneous discussion.

Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.

NVIDIA’s fiscal 2026 third-quarter filing gives a more concrete figure: approximately $50 million in H20 revenue under the relevant licenses during the filing’s disclosed period. The filing does not, in the reviewed disclosure, confirm that 15% of that amount was remitted. NVIDIA fiscal 2026 third-quarter filing

No comparable realized AMD payment amount is established in the reviewed filings. Therefore, saying that the government “received billions” or that NVIDIA paid a confirmed $7.5 million would overstate the evidence.

Rank #4
Waveshare Hailo-8 M.2 AI Accelerator Module, Compatible with Raspberry Pi 5, Supports Linux/Windows Systems, Based On The 26TOPS Hailo-8 AI Processor, Comes with PCIe to M.2 Adapter Board
  • ✅Powered by 26 Tera-Operations Per Second (TOPS) Hailo-8 AI Processor. 2.5W typical power consumption
  • ✅Scalable, enabling simultaneous processing of multi-streams & multi-models
  • ✅Enabling real-time, low latency and high-efficiency AI inferencing on the edge devices
  • ✅Supports TensorFlow, TensorFlow Lite, ONNX, Keras, Pytorch frameworks
  • ✅Supports Linux and Windows. Supports the temperature range of -40°C to 85°C

Why did Washington allow the sales?

The administration presented the policy as part of a broader U.S.–China negotiation. Commerce Secretary Howard Lutnick characterized the H20 as a lower-tier product compared with NVIDIA’s frontier accelerators. Officials also argued that allowing controlled sales could keep U.S. companies commercially present in China instead of surrendering customers and technical ecosystems to domestic suppliers.

Contemporaneous reporting connected the decision to negotiations that also involved Chinese rare-earth export restrictions. That connection was reported as part of broader bargaining; the public sources do not establish a formal one-for-one written exchange. Reuters report republished by Yahoo Finance

What’s actually slowing this PC down?

Pick the symptom - the matching free tool is one click away.

Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Support on Ko-Fi

Arguments for and against the arrangement

Why supporters backed it

  • U.S. companies retain some access to a major market rather than abandoning it entirely.
  • The government can collect money from sales that might otherwise be prohibited.
  • Continued commercial relationships may support U.S. semiconductor ecosystems and future competitiveness.
  • The administration viewed H20 and MI308 as less capable than leading U.S. accelerators.

Why critics objected

  • If the chips create a national-security risk, a payment does not reduce their usefulness to Chinese AI developers or military users.
  • Linking approval to revenue could blur export-control decisions with executive revenue collection.
  • The arrangement may create an incentive to relax controls in exchange for money.
  • China could respond by accelerating domestic alternatives or rejecting U.S. products on security and political grounds.

A Democratic letter from the House Select Committee on the Chinese Communist Party and a Senate letter raised concerns about the policy’s national-security and legal implications. House letter Senate letter

Best Value
Radxa AICore DX-M1M, 25TOPS NPU, M.2 2242 Module, Low Power Edge AI Accelerator
  • DEEPX DX-M1M NPU: Powered by the DEEPX DX-M1M neural processing unit, purpose-built for efficient on-device AI inference workloads.
  • COMPACT M.2 2242 FORM FACTOR: Fits the standard M.2 2242 slot, making it easy to integrate into embedded systems, edge devices, and compact computing platforms.
  • EDGE AI ACCELERATION: Designed to accelerate deep learning inference at the edge, enabling real-time AI applications without relying on cloud connectivity.
  • RADXA AICORE MODULE: The Radxa AICore DX-M1M delivers a plug-and-play AI compute solution ideal for robotics, smart cameras, and industrial automation.
  • WARRANTY AND ORIGIN: Backed by a 1-year manufacturer warranty and crafted with quality components for reliable long-term performance in demanding environments.

What changed in 2026?

On January 13, 2026, the Bureau of Industry and Security revised its China licensing policy for certain newer products, including NVIDIA’s H200 and AMD’s MI325X. Qualifying applications moved from a presumption of denial to case-by-case review, subject to security conditions and technical thresholds. The revised rule became effective January 15. BIS announcement Federal Register rule

NVIDIA separately reported that it received licenses for small amounts of H200 products for specified China-based customers beginning in February 2026. That disclosure does not say those H200 licenses carried the original H20/MI308 15% term. The January policy change should therefore be treated as a separate licensing development unless a specific license or filing expressly connects it to revenue sharing. NVIDIA 2026 filing

Did China actually buy the chips?

U.S. authorization, customer orders, physical shipments, recognized revenue and government payments are separate events. Chinese authorities and companies raised security concerns about H20, and later reporting described weak or stalled NVIDIA China AI-chip sales while domestic suppliers, including Huawei, gained ground. AP report on China sales and domestic competition

Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.

That means an approved license cannot be used as proof of strong demand. A company may obtain permission, fail to secure customers, ship only limited quantities, recognize modest revenue, or face payment terms that remain undisclosed.

Quick Recap

Bestseller No. 1
waveshare Hailo-8 M.2 AI Accelerator Module, Compatible with Raspberry Pi 5, Supports Linux/Windows Systems, Based On The 26TOPS Hailo-8 AI Processor, Module Only
waveshare Hailo-8 M.2 AI Accelerator Module, Compatible with Raspberry Pi 5, Supports Linux/Windows Systems, Based On The 26TOPS Hailo-8 AI Processor, Module Only
✅Scalable, enabling simultaneous processing of multi-streams & multi-models; ✅Enabling real-time, low latency and high-efficiency AI inferencing on the edge devices
$219.99
Bestseller No. 2
MX3 M.2 AI Accelerator
MX3 M.2 AI Accelerator
Software and Documentation can be accessed at the MemryX developer website
$169.00
Bestseller No. 4
Waveshare Hailo-8 M.2 AI Accelerator Module, Compatible with Raspberry Pi 5, Supports Linux/Windows Systems, Based On The 26TOPS Hailo-8 AI Processor, Comes with PCIe to M.2 Adapter Board
Waveshare Hailo-8 M.2 AI Accelerator Module, Compatible with Raspberry Pi 5, Supports Linux/Windows Systems, Based On The 26TOPS Hailo-8 AI Processor, Comes with PCIe to M.2 Adapter Board
✅Scalable, enabling simultaneous processing of multi-streams & multi-models; ✅Enabling real-time, low latency and high-efficiency AI inferencing on the edge devices
$230.99

What the headline gets wrong

  • It does not describe a tax enacted through normal tax legislation.
  • It does not mean NVIDIA and AMD paid billions; disclosed realized payment amounts are not established.
  • It does not cover all of either company’s China business.
  • It does not prove the same 15% condition applies to H200, MI325X or every later license.
  • It does not turn export approval into evidence of Chinese acceptance or large-scale shipments.

Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.

Written by TheFinanceBase Team

The Team behind TheFinanceBase.

Add your note

Your email address will not be published. Required fields are marked *

Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.

Recommended PC Tool
Recommended PC Tool
PC Slower Than It Used to Be?Free scan - under a minute
Outdated Drivers Are Slowing You DownFree scan - exact matches

Two free Windows tools

One Free Minute Could Fix That PC

Before you go - each of these free tools takes about a minute and tackles what quietly slows a Windows PC down.

Special offer. View Outbyte info, uninstall instructions, EULA, and Privacy Policy.