
A DeepSeek ban landed in the Federal Register on 20 October 2026: a Department of Homeland Security order, signed by Secretary Markwayne Mullin and dated 24 July 2026, invoking the Federal Acquisition Supply Chain Security Act of 2018, stripping DeepSeek’s artificial intelligence technology out of every US civilian agency’s own systems and barring any new federal purchase of it. Two days later, Bloomberg and the Wall Street Journal reported that DeepSeek had closed a funding round of around $15 billion, backed by Tencent, battery maker CATL and more than a dozen other Chinese investors, at a pre-money valuation of about $74 billion, up from roughly $50 billion only four months earlier. Two documents, two days apart, both about the same company, pointing in opposite directions.
The DeepSeek Ban Is Narrower Than the Headlines Suggest
The order instructs civilian agencies to remove DeepSeek products from their IT systems, along with any third-party product “developed using DeepSeek implementations,” a phrase the notice does not define, and bars new federal purchases from the company. It says nothing about DeepSeek’s commercial prospects, its revenue, or whether organisations outside US federal procurement should buy its technology, and it does not reach private companies or the public at all. China’s foreign ministry spokesperson Mao Ning called the move an example of “the overstretching of the concept of national security” at a briefing the same day the order was published.
The Market Answered a Different Question Entirely
The $15 billion round answers a question the DHS order never asked. Tencent, CATL and more than a dozen other investors reportedly including Meituan, Geely and NetEase were not voting on whether DeepSeek belongs inside a US federal network. They were pricing a Chinese AI company against its own growth, its own product roadmap, and its own prospects for a Shanghai listing, reportedly targeted as early as 2027. The valuation moved up in the same week a major government restricted its own purchasing from the company. Founder Liang Wenfeng, who retains an estimated four-fifths stake in the operating entity, did not need Washington’s approval to find backers willing to price his company higher than it was four months ago.
Two Signals, Not One Contradiction
The mistake would be treating these two facts as if they disagree with each other. The two facts reflect two separate tests, a compliance test with a specific regulatory scope, and a market test with a completely different one, and they simply returned two different results. An organisation can fail the first test entirely, by sitting squarely inside the regulatory scope a ban was written for, while the underlying vendor still passes the second comfortably, because its investors are answering a different question with different information.
What a PMO Should Actually Check
For any organisation that is not a US federal civilian agency, the DHS order carries no verdict on the vendor at all, only a compliance fact relevant to exactly the organisations it names. The question worth asking is narrower and more useful than “should we drop this vendor”: when a procurement ban lands on something your organisation uses, has anyone actually checked whether the restriction’s own wording applies to your organisation’s position, or has the ban simply been treated as proof the vendor is in trouble, because that is the easier story to tell a steering committee than reading the order itself. A vendor risk register that logs every competitor’s ban as a reason to exit, without first checking the scope of that specific ban, will eventually drop a perfectly viable supplier over a restriction that never applied to the organisation holding the register. DeepSeek’s own investors answered the viability question for themselves, and their answer was $15 billion.
Pick the last vendor decision your organisation made because of someone else’s ban. Name the regulatory scope that ban actually covered, and check whether your organisation ever sat inside it.


