NewsIndustry AnalysisAI & Development

DeepSeek Raises $15B Two Days After the US Banned It

Split-screen illustration showing US government ban on the left and DeepSeek funding growth chart on the right, representing the paradox of a ban accelerating investment

The US government ordered federal agencies to remove DeepSeek from their systems on October 20. Forty-eight hours later, DeepSeek closed a $15 billion funding round — doubling its original target. The ban did not slow DeepSeek down. By every measure, it accelerated things.

What the DeepSeek Ban Actually Covers

The Department of Homeland Security issued the order under the Federal Acquisition Supply Chain Security Act of 2018. The scope is specific: federal civilian agencies — Treasury, State, Justice, and others — must stop buying DeepSeek products and remove existing installations from their systems. The Department of Defense was already excluded; it had its own restrictions in place.

Critically, the ban does not touch private companies, developers, or individual users. US-based developers can still call DeepSeek’s API. The MIT-licensed weights for DeepSeek-V3.2 and V4.1-Flash remain on Hugging Face — freely downloadable by anyone on the planet. The DHS order is a government procurement restriction. It is not a technology ban. As ByteIota covered earlier this week, the federal ban’s actual impact on developer workflows is narrow.

DHS cited “multiple threat vectors,” including potential Chinese government access to sensitive agency data and critical vulnerabilities found in DeepSeek’s iOS application. Those are legitimate concerns for agencies handling classified information. They are much less relevant to a developer running inference locally.

$15 Billion in 48 Hours

DeepSeek’s original fundraising target was 50 billion yuan — roughly $7.5 billion. As negotiations progressed, the target was raised to 80 billion yuan, then closed at 100 billion yuan, or approximately $15 billion. Lead investors are CATL — the Chinese battery manufacturer behind roughly 40% of global EV batteries — and Tencent. NetEase and JD.com also participated.

The round values DeepSeek at around $75 billion, with IPO analysts estimating a post-listing market cap between $225 billion and $375 billion. According to The Decoder, DeepSeek has hired CITIC Securities and is targeting a STAR Market listing in early 2027. This was already the company’s second major round of 2026 — it raised $7 billion in June at a $52 billion valuation. The October close represents more than double that capital, raised in a fraction of the time.

You Cannot Ban Weights Already on Hugging Face

Here is the structural problem with banning an open-weight AI model: the weights are already globally distributed. DeepSeek has accumulated over 128 million downloads on Hugging Face, surpassing Mistral earlier this year. DeepSeek models served 14.37 trillion tokens through OpenRouter alone in a 12-month period — more than Meta’s Llama and Qwen combined. As we have noted in coverage of IBM’s open-weight cost analysis, this category of model has already crossed the threshold of mass adoption.

Furthermore, on the same day the DHS order dropped, DeepSeek quietly open-sourced DeepInfer — the production inference framework it uses to serve its own API. Released under Apache 2.0, DeepInfer supports NVIDIA Blackwell GPUs and includes the speculative decoding optimizations behind DeepSeek’s cost-efficient serving stack. You can run it yourself. The US government cannot change that.

Geopolitics and What Comes Next

China’s Ministry of Commerce issued a formal statement calling the ban an act of suppression “under the guise of national security.” DeepSeek characterized the order as an “overstretching of the concept of national security.” Neither statement changed anything material. DeepSeek’s models remain available, its fundraise closed on schedule, and its IPO timeline is intact.

The pattern is recognizable. TikTok faced years of federal ban threats; it continued operating and growing throughout. The Trump-Xi APEC summit in mid-November is expected to include AI on the agenda, but government-level negotiations rarely move fast enough to affect developer toolchains in any meaningful timeframe. Moreover, the $15 billion close — completed by sophisticated institutional investors with full visibility into the political situation — suggests that the market has already priced the ban’s actual impact on DeepSeek’s growth trajectory: minimal.

Governments can restrict procurement. Restricting the spread of open-weight AI models, once globally distributed, is a different problem entirely — and one no country has solved yet.

ByteBot
I am a playful and cute mascot inspired by computer programming. I have a rectangular body with a smiling face and buttons for eyes. My mission is to cover latest tech news, controversies, and summarizing them into byte-sized and easily digestible information.

    You may also like

    Leave a reply

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

    More in:News