DeepSeek Raises $15 Billion From Tencent and CATL — Days After the US Federal Ban

The Round: China's Largest AI Financing Ever

DeepSeek has closed a funding round of roughly 100 billion yuan — about $15 billion — the largest single-round financing in the history of China's AI industry, according to Bloomberg and multiple outlets reporting the deal this week. The Hangzhou-based lab, famous for triggering a trillion-dollar tech stock sell-off in 2025 by proving frontier-class models could be trained on a budget, is now sitting on war chest-level capital.

The most striking detail is demand. DeepSeek reportedly set an initial target of 50 billion yuan, raised it to 80 billion as investors piled in, and finally landed at 100 billion. That doesn't look like a company begging for compute money — it looks like investors competing for allocation in what they expect to be one of the decade's defining IPOs.

The Timing: A Ban on One Continent, a Buy Signal on Another

The round closed in the same week the US Department of Homeland Security ordered civilian federal agencies to remove DeepSeek from government systems, escalating the bureaucratic campaign against Chinese AI models in federal infrastructure. The two events make a neat pair: Washington treats DeepSeek as a security risk, while Tencent, CATL, and a queue of Chinese institutional money treat it as the country's best shot at a globally competitive frontier lab.

Beijing moved in the opposite direction in the same window. Following an October 16 regulatory milestone, DeepSeek and MiniMax reportedly completed AI technology export registrations and received approval for global deployment of their models — clearing the government paperwork needed to serve users overseas. The message from both capitals is the same: DeepSeek is now strategic infrastructure, not a startup curiosity.

Who Wrote the Checks

The round's anchor investors are household names in Chinese industry:

• Tencent — the WeChat operator and one of DeepSeek's largest backers, deepening its position after distribution partnerships put DeepSeek models in front of hundreds of millions of users
• CATL (Contemporary Amperex Technology) — the world's biggest battery maker, reportedly adding about $1.38 billion, roughly double its check from DeepSeek's first round
• New money — Geely, Legend Capital (Lenovo-backed), Stony Creek Capital, China Fortune-Tech Capital, and CPE Asset Management
• Returning backers — IDG Capital, Lux Capital, and Silicon 100

Per Dealroom data, the round implies a pre-money valuation around $68.8 billion, with the money split almost evenly between new and existing shareholders. Combined with a first round of more than $6.88 billion closed in June, DeepSeek has now raised roughly $20.6 billion in six months — the largest primary-market financing in China this year.

What $15 Billion Buys

DeepSeek's historical advantage was doing more with less. That era is ending by choice. The company is spending aggressively on three fronts:

• AI data centers — the compute base needed to train and serve the next generation of V-series models, reducing dependence on rented capacity
• Open-source models — continuing the MIT-licensed releases (V4, V4.1-Flash, the DeepInfer inference stack) that made DeepSeek the default cheap option in AI coding and agent tools worldwide
• An IPO — Bloomberg reports DeepSeek is preparing a listing that would be one of the most anticipated Chinese market debuts in years, with rival Moonshot on a similar track

Cheap, high-quality open-weight models have been DeepSeek's growth engine: September's V4.1-Flash release pushed share prices of domestic rivals MiniMax and Z.AI down, and its models remain fixtures inside coding assistants, research agents, and chat apps built outside China — including in markets whose governments are advising against it.

Can It Close the Gap With OpenAI and Anthropic?

Not on fundraising alone. OpenAI and Anthropic have each raised more than $100 billion, and OpenAI is reportedly nearing a $30 billion round at a $1.25 trillion valuation — though that figure is itself down from earlier $1.4 trillion estimates, a sign that even frontier-lab appetite has limits. DeepSeek's $20.6 billion is an order of magnitude smaller.

But capital efficiency has always been DeepSeek's weapon. Its sparse Mixture-of-Experts designs deliver near-frontier coding and agentic performance at a fraction of inference cost, and every dollar it now spends on data centers compounds that price advantage rather than replacing it. The gap that matters to tool builders — cost per useful task — may close faster than the headline funding gap suggests.

Upsides and Risks

What This Means for AI Tool Users

If you build with AI APIs, a funded DeepSeek means the cheapest credible model family stays credible — expect the V-series line and its Search API to keep improving while keeping pressure on OpenAI, Google, and Anthropic pricing.

If you're choosing AI tools, the apps you use that quietly route to DeepSeek under the hood (coding assistants, research agents, budget chatbots) just got a more stable foundation — even as some enterprise buyers in the US face new rules about where their model provider sits.

If you're watching the market, the IPO is the story of 2027: a Chinese frontier lab listing with tens of billions raised, export approvals in hand, and its flagship models already embedded in tools used worldwide.

You can compare DeepSeek-powered tools and hundreds of alternatives on aitrove.ai.

Frequently Asked Questions