- DeepSeek is reportedly seeking $1.5 billion at a $71 billion pre-money valuation, according to Bloomberg.
- The proposed round comes just weeks after the Chinese AI startup raised around $7 billion at a valuation of roughly $50 billion, representing a nearly 40% jump in little more than a month.
- DeepSeek is targeting an IPO by 2027, although a listing could come as early as late 2026.
DeepSeek is in talks to raise $1.5 billion in fresh capital, according toBloomberg. The new money would come on top of a $71 billion pre-money valuation, up from the roughly $50 billion valuation DeepSeek carried when it closed its first-ever external round, around $7 billion, according toReuters, which dates the close to June 16. Financial Times-sourced reporting places the same round slightly earlier, in late May, at a $52 billion post-money valuation. Either way, the new talks put DeepSeek’s valuation up close to 40% in under six weeks.
From $10 billion to $71 billion in three months
The speed only looks more extreme zoomed out.
DeepSeek first opened up to outside capital in April at a $10 billion valuation, having spent two years turning down every investor while High-Flyer, Liang’s hedge fund, funded its research alone. Within 48 hours, the target doubled past $20 billion as Tencent and Alibaba entered talks.
By May, it was nearing $45 billion, more than four times the figure on the table two weeks prior, before the round closed at $50–52 billion. Now, weeks later, DeepSeek is talking to investors about $71 billion, roughly a sevenfold increase in three months.
That climb has been personally lucrative for Liang. His stake, still around 78% after the first round diluted it, has pushed his net worth to about $36 billion, up from $16.7 billion, according to the Bloomberg Billionaires Index, making him the wealthiest AI founder globally, ahead of his counterparts at OpenAI and Anthropic.
A five-week repricing and an IPO in parallel
DeepSeek’s first outside round brought in investors including Tencent, CATL, JD.com, NetEase, and IDG Capital, as well as China’s National Artificial Intelligence Industry Investment Fund, the only investor granted direct equity and voting rights.
Every other outside investor’s capital went into a limited partnership that Liang controls, with no voting rights and a five-year lock-up.
Bloomberg reports the new round is running in parallel with early IPO preparations: DeepSeek is working with accounting and banking advisers to finalise its financial statements by the end of December, a required step before any filing.
A listing could come as soon as this year, with a debut targeted for 2027, a timeline that would put DeepSeek ahead of OpenAI, which recently pushed its own IPO to 2027 while holding out for a $1 trillion valuation. No exchange has been confirmed; the market widely expects mainland China or Hong Kong, where rivals Zhipu and MiniMax have already listed, with Zhipu’s shares up nearly 1,600% since its January debut.
That precedent may be part of what’s pulling DeepSeek toward the public markets sooner rather than later. TK: on-record quote from Liang Wenfeng or an investor on the new round.
Why investors are still paying up
DeepSeek isn’t repricing on hype alone. It accounted for nearly 23% of enterprise AI gateway tokens processed in June, according to TechCrunch’s report on Vercel’s data, second only to Anthropic’s 32% share. It built that position on open-source reasoning models that perform within striking distance of frontier US labs, despite US export restrictions limiting China’s access to advanced Nvidia chips.
DeepSeek has instead leaned on Huawei’s domestic hardware, an approach that’s become one of the clearest proof points yet that China can run a competitive AI stack without American chips.
The market backdrop helps explain the appetite. Precedence Research puts the global large language model market at $7.77 billion in 2025, projecting it to reach $149.89 billion by 2035 at a 34.44% compound annual growth rate. DeepSeek is competing directly with OpenAI, Anthropic, and xAI, plus domestic rivals Moonshot AI, MiniMax, and Zhipu AI, all raising multi-billion-dollar rounds of their own.
What differentiates DeepSeek is that it’s pairing open-source distribution with domestic infrastructure rather than a closed, capital-intensive stack. Whether that holds up once DeepSeek is a public company, answerable to shareholders instead of a five-year-locked-up private cap table, is the real question the IPO will test.
