Chinese AI lab DeepSeek is closing in on roughly $7.4 billion in its first ever outside funding round, a deal that would value the open-source model maker at between $52 billion and $59 billion and rank among the largest private technology financings in China's history.
Bloomberg and Reuters report that Tencent and battery giant CATL have begun signing term sheets to anchor the round, with Beijing's state-backed National Artificial Intelligence Industry Investment Fund also taking part. Founder Liang Wenfeng is expected to contribute roughly 40% of the total himself, sources told the wires on Wednesday.
Why the round matters
DeepSeek became globally famous after releasing its V3 and R1 models without taking a dime of outside venture capital. The cost-efficient training approach behind those models rattled US chip stocks last year and pushed the company to the front of China's AI race. The current cash call signals that even DeepSeek can no longer self-fund the inference clusters, agentic frameworks and V5 training runs the next phase will require.
Tencent is reportedly putting in 10 billion yuan and CATL another 5 billion yuan, making them DeepSeek's two largest external backers. Late-stage talks are also under way with NetEase, JD.com, Hong Kong-based IDG Capital and Monolith Management. The China National AI fund rounds out a uniquely state-flavoured cap table.
A different playbook from OpenAI and Anthropic
Even at $59 billion, DeepSeek would still be a fraction of the size of US frontier labs. OpenAI reportedly raised $122 billion earlier in 2026, and Anthropic banked $65 billion in May. With US export restrictions cutting Chinese labs off from Nvidia's most advanced GPUs, DeepSeek has chosen breadth – open-source releases, domestic chip partners and energy-aware infrastructure – over brute compute.
That strategy is already drawing follow-on coverage from peers. Earlier this year, observability vendor Coralogix raised $200 million to police the same AI-agent workloads DeepSeek wants to power, while Tempus AI's tumor-only diagnostic clearance shows enterprises are quickly operationalising the models DeepSeek and rivals release.
CATL's not-so-quiet pivot
CATL's involvement is the most interesting industrial signal. The company dominates EV batteries, but it has been steadily building energy-storage and power infrastructure for hyperscale data centers, and even hardware partnerships across China's AI hardware supply chain. Putting capital into DeepSeek effectively gives CATL a seat at the table for the next decade of AI energy demand – mirroring deals like NVIDIA's tighter alignment with TSMC across the AI supply chain.
What happens next
People familiar with the deal said DeepSeek expects to close the round within weeks. There is no IPO timeline, though the structure – with strategic corporates, a state fund and the founder personally on the cap table – is closer to a Chinese national-champion blueprint than a Silicon Valley unicorn round. The capital is expected to fund compute access, a planned V5 model and a push into AI agents that DeepSeek has signalled since its April release of V4.
Reporting based on coverage from Bloomberg, Reuters, Tech Startups and the South China Morning Post.