• China’s Commerce Ministry said the U.S. order excluding DeepSeek from federal systems disregards consensus reached between Chinese and U.S. heads of state.
  • The Department of Homeland Security made the order public on October 19, with Federal Register publication set for October 20; the underlying order was signed on July 24, 2026.
  • The restriction applies only to federal civilian agencies, not to private businesses or consumers, but it extends to third-party products built using DeepSeek technology.

Escalating Tech Dispute

China’s Commerce Ministry on Tuesday accused the United States of disregarding a consensus reached between the two countries’ leaders, after Washington moved to bar DeepSeek’s artificial intelligence technology from federal civilian government systems. The rebuke marks a fresh escalation in a widening technology and trade dispute between the world’s two largest economies.

The U.S. Department of Homeland Security made the order public on October 19, with publication in the Federal Register scheduled for October 20. The underlying order was signed on July 24, 2026, according to people familiar with the matter, meaning the latest development is its public disclosure rather than a newly signed prohibition.

Scope of the Order

Under the order, federal civilian agencies must remove DeepSeek and affiliated products, including third-party products and services developed using DeepSeek technology, and stop procuring covered technology. The Department of Defense is excluded and operates under separate restrictions. Agencies can request exceptions.

DHS cited potential transfers of sensitive government information to China, possible Chinese government access to company infrastructure, and security vulnerabilities. These are the U.S. government’s stated risk assessments; reporting has not established that a particular federal data breach occurred.

A Chinese Foreign Ministry spokesperson, Mao Ning, said Beijing would protect Chinese companies’ legitimate rights and opposed what she called the overstretching of the concept of national security and the politicization of technology and trade. DeepSeek did not immediately respond to a request for comment.

DeepSeek’s Financial Position

DeepSeek is a privately held Chinese AI company based in Hangzhou, led by founder and CEO Liang Wenfeng. Its business includes general-purpose and reasoning models, a chatbot, and developer access to models. Its latest model identified in reviewed financial reporting is DeepSeek-V4.1-Flash, released in September 2026.

CNBC reported a June external financing of approximately 50 billion yuan, at an implied valuation of roughly 350 billion yuan. Reuters reported on September 24, citing The Information, that annualized revenue run rate had reached $1 billion—more than twice the level a few months earlier—though Reuters explicitly said it could not independently verify that report. A run rate extrapolates recent revenue; it is not actual full-year revenue or profit.

The same report attributed some revenue growth to model-price increases of 2.3–4.5 times, so the increase should not be read purely as growth in customer usage. CNBC reported October 6 that DeepSeek was considering expanding its latest round to as much as 100 billion yuan, approximately $14.9 billion, at a targeted valuation near $75 billion. The amount remained fluid and was not a confirmed completed financing.

Liang remains founder and CEO; reviewed sources report no confirmed CEO change or major operational restructuring. The significant corporate shift is toward outside funding and preparations for a possible Shanghai STAR Market listing, with CITIC Securities (600030.SS) assisting. IPO timing and terms remain unsettled.

Market and Industry Implications

The most direct exposure is among federal agencies and their technology suppliers. Because the removal requirement extends to third-party implementations, vendors may need to identify embedded DeepSeek components, replace models, and retest applications. The implementation burden follows from the order’s scope; reviewed sources do not quantify its total cost.

For DeepSeek, the potential damage extends beyond lost government sales to reputational and procurement risk elsewhere. However, sources do not disclose its U.S. federal revenue exposure, so a material earnings impact cannot yet be calculated.

DeepSeek illustrates two competing forces in AI: lower-cost competition and rising capital intensity. Its January 2025 models challenged assumptions about the cost of advanced AI and triggered a U.S. technology-stock sell-off. That history explains why restrictions on the company attract attention beyond the relatively narrow government market.

The order uses the Federal Acquisition Supply Chain Security Act of 2018 framework. That matters because Washington is treating DeepSeek as a government supply-chain security issue, rather than imposing a universal consumer-service prohibition.

Political and Diplomatic Context

Beijing’s “leaders’ consensus” framing presents the dispute as a challenge to bilateral cooperation. Without the specific agreement or transcript, however, it would be premature to conclude that Washington breached a defined commitment.

The diplomatic timing is sensitive. Presidents Donald Trump and Xi Jinping are expected to meet in mid-November at the APEC summit in Shenzhen, with AI likely on the agenda. The two countries have also made progress toward communication arrangements for national-security-related AI incidents. Restrictions and limited safety cooperation are therefore proceeding simultaneously.

Other tensions remain unresolved: U.S.-led controls restrict Chinese access to advanced AI chips and chipmaking equipment, while Beijing proposed restrictions on exports of certain advanced AI models the previous week. These are separate policies, but they reinforce the struggle over who controls AI infrastructure and capabilities.

What to Watch

The practical issue is implementation: identifying affected third-party software, replacing models, and determining which agencies qualify for exceptions. Technology publication The Register highlights the difficulty of identifying every product developed using DeepSeek—a more immediate operational concern than the headline’s diplomatic language.

For investors, the key unresolved question is whether the restriction affects DeepSeek’s fundraising terms or prospective listing. Its financing discussions were already fluid before this disclosure; there is no verified evidence in reviewed sources that the order has derailed them.

A plausible outcome is further separation between U.S. government AI procurement and Chinese-developed models, with model provenance becoming a standard compliance consideration. The extent of spillover into private-sector restrictions remains uncertain; this order alone does not establish such a prohibition.

The November leaders’ meeting will be a useful test of whether both governments can preserve limited AI-safety cooperation while continuing to restrict each other’s technology. AI is expected to be a discussion topic, but no verified negotiated resolution to this DeepSeek dispute has been announced.

Update: This article was updated to clarify that the order applies to federal civilian agencies and excludes the Department of Defense.