DeepSeek Closes ~$15B Round with CATL and Tencent: What It Means for Your Business AI Costs | inicIA
Direct answer
DeepSeek is closing a round of ~100 billion yuan (≈$15 billion) from new and existing investors, according to Reuters (Oct 22, 2026) citing four sources close to the process. The biggest investors would be CATL (the battery giant) and Tencent, alongside Midea, Meituan, Geely and GigaDevice as new participants, plus Boyu Capital, CPE, Legend Capital and IDG Capital. The round doubled the original 50B yuan target and values the company at 500B yuan pre-money (previously reported by Reuters; Bloomberg broke the news on Oct 22). The signal for your business: the race for powerful, cheap models is not stopping, and that keeps pushing AI costs down.
The Round's Numbers
- ~$15B raised (≈100B yuan), double the initial 50B yuan target: investor appetite for Chinese AI is real, with 15+ investors in the round.
- Who's writing the checks: CATL and Tencent lead; Midea (appliances), Meituan (delivery), Geely (autos) and GigaDevice (semiconductors) join. It's Chinese industrial and state-backed capital betting on the race against U.S. labs.
- 500B yuan pre-money valuation (≈$75B): DeepSeek consolidates as the most valuable open lab outside the U.S.
- The context: it lands days after the U.S. ordered federal agencies to remove DeepSeek over security risks (DHS, Oct 19). The market is splitting in two: massive global adoption and U.S. government veto.
Why This Lowers Your AI Costs
- Cheap open models are your negotiating lever: every round funding a low-cost competitor (DeepSeek, Qwen, Kolibri) pressures the API prices your business pays. We already saw this in 2026 with the price war between OpenAI, Anthropic and Google.
- Evaluate open models for non-sensitive workloads: if your use case doesn't touch regulated data or trade secrets, a self-hosted open model can cost a fraction of a frontier API. DeepSeek's round funds exactly that alternative.
- Don't marry a single vendor: design your stack to switch models without rewriting everything (the same "AI of choice" logic). The 2027 price map will be drawn by those raising capital today.
The Risk You Can't Ignore
- Geopolitical and compliance risk: the U.S. already banned DeepSeek in federal agencies and flagged it as a threat vector. If you operate in regulated sectors or handle sensitive data, a Chinese model can close commercial doors or create legal exposure.
- The round isn't fully closed: Reuters cautions that financial details and the final investor lineup could still change. Don't make architecture decisions on moving headlines.
Related links
- SMB AI spending soared and Anthropic dethroned OpenAI: your advantage as a buyer
- The war of cheap verdicts: the AI layer that decides "yes or no" without writing an essay
- Claude Haiku 5.5 costs up to 90% less: the AI price war continues (and benefits you)
Frequently asked questions
How much did DeepSeek raise in this round?
Around 100 billion yuan (≈$15 billion), per Reuters (Oct 22, 2026), double its original 50B yuan target. The round was still closing and final details could change.
Who are the main investors?
CATL (batteries) and Tencent (WeChat) would be the biggest investors, with Midea, Meituan, Geely and GigaDevice as new participants, plus Boyu Capital, CPE, Legend Capital and IDG Capital.
Why does it matter for my business if I don't use DeepSeek?
Because it funds cheap, open-model competition, which pressures all AI API prices down. It's the same dynamic that cheapened frontier models during 2026.
Should my company use DeepSeek?
It depends on your risk profile: it's powerful and cheap, but the U.S. vetoed it in federal agencies over security. In regulated sectors or with sensitive data, weigh compliance risk before the savings.
