DeepSeek raises 15 billion in a reported funding round that arrived just two days after the United States ordered federal civilian agencies to remove its artificial intelligence technology from their systems. The timing is what makes this story travel: in Washington, DeepSeek was being treated as a security risk to be stripped out of government computers; in China, investors were lining up to put more money behind it.
The number needs one important word attached to it: reportedly. Bloomberg first reported the round, TechCrunch picked it up on October 22, and DeepSeek did not immediately respond to a request for comment. So this is not a company announcement with a signed term sheet attached. It is a large, well-sourced report — and the distinction matters, because the size of the round is the story.
If the report is right, about $15 billion, or roughly 100 billion yuan, would make this one of the biggest funding rounds in China’s history. That is not a rescue round for a company in trouble. It looks, from the outside, like a bet that DeepSeek’s future will be decided far more by Chinese capital, Chinese infrastructure and Chinese users than by access to American government contracts it never really had.
Key takeaways
- DeepSeek reportedly raised about $15 billion, or about 100 billion yuan, Bloomberg reported via TechCrunch on October 22, 2026.
- The report came two days after the US Department of Homeland Security ordered federal civilian agencies to remove DeepSeek products from their IT systems.
- Potential investors named in the report include battery maker CATL and Tencent, the company behind WeChat.
- The round follows a June raise of more than $7 billion at a $50 billion valuation, and a reported July effort to raise a further $1.5 billion at around $71 billion.
- DeepSeek is reportedly planning to spend more on AI infrastructure and open models, for domestic and international markets, despite US restrictions.
- A potential IPO in China as early as next year has been reported, but no listing has been confirmed.
- DeepSeek did not immediately respond to TechCrunch’s request for comment, so the final size and investor list remain unconfirmed.
Two days, two very different messages
On October 19, the Department of Homeland Security published notice of an order excluding DeepSeek products from federal civilian agency procurements and removing DeepSeek products — including third-party products built on DeepSeek implementations — from civilian agency IT systems. The underlying order had been signed in July, but its public appearance in October is what forced agencies, vendors and the press to deal with it all at once.
Two days later, on October 22, came the funding report.
Put side by side, those two events tell a story about how split the global AI market has become. One government can remove a model from its own systems. It cannot, on its own, remove that model from the market, from open-weight downloads, from Chinese data centres, or from the balance sheets of investors who see a domestic champion worth backing.
That does not mean the US order is meaningless. For American agencies and their suppliers, it is a hard procurement rule with real compliance work behind it.
What the US order actually does — and does not do
It helps to be precise here, because “ban” is used loosely in headlines.
The DHS order applies to US federal civilian executive branch agencies. It bars them from acquiring DeepSeek products and services, including those from affiliated entities, and requires removal of DeepSeek products already in their systems. Crucially, it also reaches third-party products and services developed using DeepSeek implementations. That is the part that creates work for vendors: an agency cannot simply check whether it bought the DeepSeek app. It has to ask whether a tool it uses was built on top of DeepSeek’s models.
The order does not, by itself, ban private American companies or ordinary users from using DeepSeek. It is a federal supply-chain security measure, made under the Federal Acquisition Supply Chain Security Act of 2018, following a review by the Federal Acquisition Security Council. DHS said DeepSeek’s models present “multiple threat vectors” and warned that use “may result in the transfer of sensitive data to the PRC if used by U.S. government employees or vendors.” It also said use could produce actionable intelligence on US persons or agencies, and that the models themselves may contain malware or other security vulnerabilities.
Those are the government’s stated concerns, not proven findings about every DeepSeek product. The order is precautionary in design: it removes a perceived risk from government systems rather than waiting for proof of a specific incident. Agencies can seek exceptions, but they need a compelling justification, and reporting on the order noted that exception requests were supposed to be submitted 60 days in advance — a detail that, in practice, makes exceptions hard to obtain after the fact.
Some US agencies and states had already restricted DeepSeek before this order. What changed in October was the scale: a government-wide civilian rule, rather than a series of individual agency decisions.
Why investors would still write a cheque this big
A $15 billion round, even a reported one, is not a casual vote of confidence. It is a statement about where the investors think value will be created.
First, DeepSeek does not need the US federal market to be a large company. Its home market is China, where it is based in Hangzhou, and where demand for domestic AI models, cloud capacity and developer tools is enormous. A Chinese battery giant like CATL and a platform company like Tencent are not buying access to Washington. They are buying exposure to a model maker that could sit underneath a wide range of Chinese industrial and consumer applications.
Second, DeepSeek’s model strategy is different from a closed, subscription-only lab. It gained global attention in early 2025 with a free-to-use, open-weight reasoning model that challenged the dominance of US AI companies. Open models travel differently from closed ones: developers can run, adapt and build on them, even when a particular government procurement door closes. That does not make restrictions irrelevant, but it does make them less fatal than they would be for a company whose entire business was selling to one government.
Third, the funding history shows momentum, not a one-off spike. In June, DeepSeek raised more than $7 billion at a $50 billion valuation. In July, it was reportedly seeking a further $1.5 billion at a valuation of around $71 billion. A $15 billion round on top of that, if confirmed, would be a step-change in scale — the kind of money that buys data centres, chips, talent and time, rather than just another few months of runway.
That scale is also the point for rivals. TechCrunch’s report notes that DeepSeek is ramping up spending on AI infrastructure and open models for domestic and international markets, and that its expansion could intensify competition among domestic rivals. In other words, this is not just DeepSeek versus OpenAI or Anthropic. It is DeepSeek versus every other Chinese lab fighting for the same engineers, the same compute and the same enterprise customers.
The wider Chinese AI funding wave
DeepSeek is the headline, but it is not happening in isolation.
Moonshot AI, the developer of Kimi, secured $2 billion at a $20 billion valuation in May. Manus AI raised more than $500 million earlier in October. Zhipu AI (Z.ai) and MiniMax both went public in Hong Kong in January, and had market capitalizations of roughly $29 billion and $7.2 billion respectively as of October 22, according to TechCrunch’s report.
Read together, those numbers describe a market where capital is not drying up in the face of geopolitical tension. It is reorganising around it. Chinese investors, industrial companies and public markets are funding a domestic AI stack — models, agents, infrastructure and applications — that is designed to survive without American government approval, and in some cases without American capital markets at all.
That reorganisation has a cost. Building a separate AI ecosystem means duplicating chips, software, standards and talent pipelines. It also means Chinese labs will face their own version of the question US labs face: how to pay for enormous infrastructure spending before the revenue catches up. A $15 billion round answers that question for a while. It does not answer it forever.
What this means for US labs and for ordinary users
For American AI companies, the DeepSeek round is a reminder that restrictions can slow a rival in one market while accelerating it in another. A model that is removed from US federal systems can still be downloaded, studied, improved and deployed elsewhere. Open-weight models, in particular, are hard to contain with procurement rules alone.
That does not mean US labs should panic. They still lead on many benchmarks, on enterprise adoption and on access to the most advanced chips. But the gap that looked unbridgeable in early 2025, when DeepSeek first shocked the market, has not stayed that wide — and a $15 billion war chest will not make it wider in America’s favour.
For ordinary users, the practical effect is more mixed. If you use DeepSeek’s consumer app or a third-party tool built on its models, the US federal order does not directly change your access. What may change is how vendors talk about their supply chains: companies that sell to the US government will have to prove they are not quietly running on DeepSeek underneath, and some may drop DeepSeek-based tools to avoid the compliance risk. That could make the AI market feel more fragmented — different tools, different models, different rules, depending on who you are and who you sell to.
For developers, the lesson is about provenance. It is no longer enough to know that a tool works. Increasingly, buyers — especially government buyers — want to know what model it was built on, where its data goes, and who can demand access to it. DeepSeek’s open-model approach makes that question both easier to ask and harder to answer simply, because the same weights can be hosted in many places, under many different operators.
What remains unknown
A story this large, reported this quickly, leaves several questions open, and a careful reader should hold them in mind.
First, the round itself: who actually invested, how much each put in, and at what valuation? “Potential investors include CATL and Tencent, among others” is not a cap table. Until DeepSeek or its investors confirm details, the $15 billion should be treated as a reported total, not a closed fact.
Second, the IPO: TechCrunch reports plans for a potential IPO in China as early as next year. That is a plan, not a filing, and market conditions, regulatory approval and the company’s own readiness could all move the date.
Third, the technology: what will this money actually build? “AI infrastructure and open models” is a broad brief. The interesting detail will be whether DeepSeek prioritises larger reasoning models, cheaper inference, agent tools, or something else entirely — and whether its next release narrows or widens the gap with US frontier labs.
Fourth, the politics: will other governments follow the US civilian-agency approach, or will they treat DeepSeek as a commercial model like any other? A single US procurement order is a signal. A pattern of similar orders across allied governments would be a different kind of pressure.
None of those unknowns makes the funding report unimportant. They make it important to report it accurately: as a major, reported, unconfirmed round, at a moment when the US and China are pulling the AI market in opposite directions.
What happens next
- Confirmation — or correction. Watch for DeepSeek, CATL or Tencent to confirm, clarify or stay silent. Silence would not disprove the report, but it would leave the exact size and investor list open.
- The IPO timeline. If a China listing is genuinely planned for next year, preparatory steps — restructuring, advisers, regulatory engagement — should start to leak out in the coming months.
- A new model or infrastructure announcement. Large rounds are usually followed, sooner or later, by a product that justifies them. DeepSeek’s next major release will be read as a test of whether the money is buying capability, not just capacity.
- More supply-chain rules. The DHS order’s reach into third-party tools is likely to be copied, challenged or clarified. Vendors selling to governments will need clearer guidance on what counts as a “DeepSeek implementation.”
- The domestic race in China. With Moonshot, Manus, Zhipu and MiniMax all raising or listing, DeepSeek’s new funding raises the bar for everyone else. Expect more consolidation, more partnerships with industrial giants, and more pressure on smaller labs that cannot match this scale.
For now, the simplest summary holds: Washington told its agencies to take DeepSeek out. Investors, reportedly, just put $15 billion in. Both things can be true — and the tension between them is the story of the global AI race in 2026.
FAQs
How much did DeepSeek reportedly raise? About $15 billion, or about 100 billion yuan, according to Bloomberg as reported by TechCrunch on October 22, 2026. The round has not been confirmed by DeepSeek itself.
Why is the timing significant? The funding report came just two days after the US Department of Homeland Security ordered federal civilian agencies to remove DeepSeek products from their IT systems over security concerns.
Who are the reported investors? Potential investors named in the report include Chinese battery maker CATL and Tencent, the company behind WeChat, among others. A full investor list has not been confirmed.
What did the US order actually ban? It barred US federal civilian agencies from acquiring DeepSeek products and services, and required removal of DeepSeek products — including third-party products built on DeepSeek implementations — from civilian agency IT systems. It does not, by itself, ban private companies or ordinary users from using DeepSeek.
Why did the US say it acted? DHS said DeepSeek’s models present “multiple threat vectors,” including the risk that sensitive data could be transferred to China if used by government employees or vendors, and that the models may contain malware or other security vulnerabilities. These are the government’s stated concerns in a precautionary supply-chain order.
What had DeepSeek raised before this round? More than $7 billion in June at a $50 billion valuation, according to TechCrunch, with a reported July effort to raise a further $1.5 billion at around $71 billion.
Is DeepSeek planning an IPO? A potential IPO in China as early as next year has been reported, but no listing has been confirmed.
Where will the money go? According to the report, toward AI infrastructure and open models for domestic and international markets, even as DeepSeek faces US restrictions.
Sources: TechCrunch; Bloomberg reporting as cited by TechCrunch; US Department of Homeland Security exclusion and removal order, public notice Oct. 19, 2026. Editor note: The $15 billion figure is reported, not company-confirmed. DeepSeek did not immediately respond to TechCrunch’s request for comment. Treat investor names and the IPO timeline as reported, not confirmed, until DeepSeek or the named investors say otherwise.
DeepSeek raises 15 billion: what to watch next
The reported round will be judged by what follows it: confirmation of the investors and size, any IPO steps in China, and DeepSeek’s next major model or infrastructure announcement. For now, the reported $15 billion round stands as a major test of how far Chinese AI funding can go while US federal restrictions remain in place. More coverage is available on World Reach Daily.