DeepSeek Nears $12B Raise, Leads $400M-Valuation Robotics Bet on Winverse AI

A $12 billion war chest meets the robot race

DeepSeek, the Chinese AI lab known for lean, efficient open-weight models, is close to raising at least 80 billion yuan — roughly $12 billion — in a funding round backed by investors including Tencent and battery giant CATL, Bloomberg reported in early October. The raise reportedly exceeded the company's initial target.

And the money is already moving. On October 12, Bloomberg reported that DeepSeek agreed to lead a seed round worth tens of millions of dollars for Winverse AI, a stealth-mode robotics startup founded by a former top Tencent AI scientist, at a post-money valuation of about $400 million. Meituan's investment arm is participating alongside DeepSeek.

Why this matters beyond one investment

This is the open-weight playbook maturing. DeepSeek's efficient models pressured competitors to cut prices and rethink compute strategy. Now the lab is converting that momentum into capital and deploying it into embodied AI — robots — the next frontier every major lab is racing toward.

It also sharpens the US-China AI competition. Open-weight models let anyone download parameters and adapt systems locally, and Chinese developers are using that openness to compete with US providers without matching their spending. Cheaper, efficient models plus physical AI is a combination that could compress timelines across the industry.

For the broader market, a $12 billion raise into a Chinese AI lab — in the same year Anthropic raised $30 billion and OpenAI raised $122 billion — shows capital is still flowing aggressively to frontier model companies on both sides of the Pacific.

What this means for your business

1. Open-weight models are a strategic option, not just a research toy. If you need private deployment, local customization, or freedom from a single vendor's pricing, evaluate open-weight models seriously in your next AI build-vs-buy decision. 2. Embodied AI is entering the investment mainstream. DeepSeek, Mecka AI, and XPeng's Dogotix are all pulling big checks into robotics this month. If your business touches logistics, manufacturing, or field operations, the physical-AI wave is worth tracking now, not in 2028. 3. Model costs keep falling. DeepSeek's efficiency-first approach keeps dragging inference prices down. Renegotiate AI contracts annually — what was expensive last year is cheap this year.

The bottom line: DeepSeek is no longer just an efficient model shop. With a $12B raise and a robotics bet, it is positioning as a full-stack challenger — and its efficiency playbook keeps making AI cheaper for everyone else.