DeepSeek just pulled off one of the biggest private fundraises in Chinese history. The Hangzhou AI lab is wrapping up a round of roughly 100 billion yuan — about $14.98 billion — Bloomberg reported on Thursday, October 22, 2026, citing people familiar with the matter. DeepSeek’s $15B funding round would more than double the company’s own original target and put China’s most efficient AI lab on a war footing against OpenAI and Anthropic.
That target didn’t just get met. DeepSeek raised it at least twice during talks — from 50 billion yuan to 80 billion, then to the final figure — a classic sign that more money wanted in than the company planned to take. Here is what was reported, who is backing it, and what it means for anyone building on AI.
What was actually reported
The details come from people who spoke on condition of anonymity because the deal is not public. Reuters could not immediately verify the report, and DeepSeek did not respond to requests for comment — so treat the numbers as reported, not confirmed. That said, four separate sources confirmed the broad strokes to Reuters, and Bloomberg broke the story first.
| Detail | Reported figure |
|---|---|
| Amount raised | ~100 billion yuan (≈ $14.98B) |
| Biggest backers | CATL (Contemporary Amperex Technology) · Tencent |
| New investors | Midea, Meituan, Geely, GigaDevice |
| Also participating | Boyu Capital, CPE, Legend Capital, IDG Capital (existing) |
| Total investors | More than 15 |
| Reported valuation | 500 billion yuan, pre-money |
| Exchange rate used | $1 = 6.6749 yuan |
Who is buying in — and what the lineup signals
The investor roster reads like a who’s who of Chinese industry: battery giant CATL and WeChat operator Tencent are the anchor backers, joined by appliance maker Midea, food-delivery giant Meituan, automaker Geely, and semiconductor maker GigaDevice. Financial investors Boyu Capital, CPE, and Legend Capital are in, alongside existing shareholder IDG Capital.
Two signals stand out:
- Corporate China is funding the AI race directly. This isn’t just VC money — it’s the battery, EV, delivery, and chip giants writing checks. That industrial alignment is exactly how China builds national champions.
- Demand exceeded supply. Raising the target twice (50B → 80B → 100B yuan) means investor demand outran DeepSeek’s own plans — even as Reuters reports a global tech selloff this month over AI spending fears, with Hong Kong-listed AI labs Z.AI and MiniMax down sharply from their peaks.
June vs. October: DeepSeek’s war chest, doubled in 4 months
This is DeepSeek’s second external raise of the year — and the pace is striking:
| June 2026 | October 2026 (reported) | |
|---|---|---|
| Raised | $7.4 billion | ≈ $15 billion |
| Implied valuation | 350.88B yuan (≈ $52B, per investor filings) | 500B yuan, pre-money |
| Significance | First external round | Among China’s largest private raises ever |
DeepSeek’s 2026 funding arc
June 2026 — $7.4B
First external raise. Investor filings pointed to a ~350.88B yuan valuation (≈ $52B).
September 2026
Reuters reported DeepSeek aimed to kick off its IPO process this year, hiring CITIC Securities for a STAR Market float.
October 2026 — ≈ $15B (reported)
Target raised twice to 100B yuan. 500B yuan pre-money valuation. 15+ investors, CATL and Tencent anchoring.
Why this is a U.S.-vs.-China story, not just a funding story
The timing is impossible to ignore. The raise lands days after Washington ordered DeepSeek removed from federal civilian systems — a direct escalation in the AI decoupling fight. DeepSeek’s answer, funded by this round: build compute independence.
That answer is already visible. Earlier this week, DeepSeek open-sourced DeepInfer, its production inference engine — the stack its README says powers its large-scale production services, supporting both NVIDIA chips and Huawei’s Ascend 950. The $15B war chest funds exactly this: domestic hardware paths, massive training clusters, and the talent to run them. And last month, DeepSeek released DeepSeek-V4.1-Flash, designed for greater capability, faster inference, higher throughput, and scaling to larger models — the product line this money will extend.
The macro backdrop cuts both ways. Reuters notes the raise comes amid a fresh selloff in global tech stock over massive AI spending, with data-center IPOs delayed or withdrawn and Hong Kong-listed Z.AI and MiniMax off their peaks. Money is getting pickier — which makes DeepSeek’s oversubscribed round more notable, not less.
How to profit: what the $15B means for your AI budget
Big war chests at efficient labs historically mean one thing for builders: stable, cheap inference for a long time. DeepSeek already competes on price, and this raise removes any near-term pressure to hike it. Concrete moves:
- Build margin-arbitrage tools on DeepSeek’s API now. Services priced against $20/month Claude or GPT subscriptions but delivered on DeepSeek inference keep the spread. We covered DeepSeek’s recent pricing moves in DeepSeek Prices Spike 230%: 17 AI Models Quietly Hiked — watch the official DeepSeek API changelog for the next move.
- Self-host with DeepInfer. DeepSeek’s open-source inference stack runs on NVIDIA and Huawei Ascend 950 hardware — a real path to zero per-token cost at volume.
- Segment your compliance story. If you sell to U.S. government-adjacent clients, DeepSeek’s federal removal matters — document your model supply chain now, before procurement asks.
- Track the IPO. A STAR Market listing would make DeepSeek’s numbers public for the first time — API costs, margins, token volumes — all data you can trade on.
Keep Reading
FAQ
How much did DeepSeek raise?
About 100 billion yuan (≈ $14.98 billion), Bloomberg reported on October 22, 2026, citing people familiar with the matter. Four sources confirmed the broad strokes to Reuters.
Who are the biggest investors in DeepSeek’s round?
Battery maker CATL (Contemporary Amperex Technology) and Tencent are among the largest backers, per two sources. New investors include Midea, Meituan, Geely, and GigaDevice.
What is DeepSeek’s valuation after this round?
Reuters reported a 500 billion yuan pre-money valuation. That is up from an implied ~350.88 billion yuan (≈ $52B) suggested by investor filings after the June $7.4B round.
Is the $15B raise confirmed?
Not officially. The figures come from Bloomberg’s and Reuters’ sources; Reuters could not immediately verify them, and DeepSeek, CATL, and Tencent did not respond to comment requests.
Is DeepSeek planning an IPO?
Reuters reported in September 2026 that DeepSeek aimed to kick off its IPO process this year and had hired CITIC Securities to prepare a potential STAR Market float.
Why does the raise matter for AI prices?
A deep war chest at the industry’s most cost-efficient lab reduces pressure to raise API prices — good news for builders arbitraging cheap DeepSeek inference against premium-priced competitors.
What did DeepSeek release most recently?
Last month it launched DeepSeek-V4.1-Flash (faster inference, higher throughput), and this week it open-sourced DeepInfer, its production inference engine supporting NVIDIA and Huawei Ascend 950 chips.
Sources
- Reuters: DeepSeek wrapping up $15 billion fundraising to fuel AI race
- Reuters: DeepSeek $15B round, sources say (investor lineup and valuation)
- AI Parabellum: DeepSeek funding round brings in about $15 billion
- DeepInfer — DeepSeek’s official open-source inference engine (GitHub)
- DeepSeek API changelog (official)