DeepSeek has started preparing to go public. The Hangzhou AI developer is working with accounting and banking advisors on a mainland China listing, with a filing planned for late this year or early 2027 and a debut targeted for 2027, according to Bloomberg reporting. At the same time, it is out raising fresh money at roughly a $71 billion valuation.
The money, and what's odd about the timing
Here's the part that stands out. DeepSeek closed its first-ever outside funding round barely a month ago, $7 billion at around a $50 billion valuation, and it's already back in the market. The new round targets about $1.5 billion, and the valuation has jumped to roughly $71 billion (480 billion yuan, per the reporting). A 40% markup in a few weeks, on a company that keeps telling investors it isn't chasing revenue.
That last bit matters. Liang Wenfeng has reportedly told backers he wants to keep shipping open-source models and stay pointed at AGI rather than commercialize what already exists. It's a nice story for a research lab. It's a stranger one for a company that also wants public-market investors to hand it a $71 billion price tag in two years.
Where the cash goes
Compute, mostly. The new capital is expected to fund a build-out of computing capacity, which is the same money pit every frontier lab is currently shoveling cash into. DeepSeek's whole reputation, going back to the V3 and R1 models that rattled Silicon Valley in early 2025, was doing more with less. Raising billions for data centers is a funny sequel to that pitch, though it's also just what building models at this scale now costs.
The company reportedly needs its financials cleaned up by end of December to file, which is what's setting the timeline.
Founder now worth $36 billion
The June round roughly doubled Liang's paper wealth, from about $16.7 billion to $36 billion, per the Billionaires Index. That reportedly puts him ahead of Anthropic's Dario Amodei and OpenAI's Greg Brockman, which is a strange sentence given DeepSeek is younger than both companies and made its name partly by undercutting exactly the kind of spending those labs are famous for.
Not the only AI lab racing to list
Anthropic and OpenAI both filed confidential IPO paperwork in June. OpenAI's CFO has floated waiting until 2027, citing cash burn and the general misery of public reporting. So DeepSeek isn't moving in a vacuum. What makes it notable is the venue: a mainland listing, likely Shanghai's STAR Market, rather than the New York route that's off the table for a Chinese AI firm anyway.
Plans are described as fluid, and both the funding and the filing could slip. The next real marker is December, when DeepSeek needs its financial statements done to file at all.
