DeepSeek Raises Over 50 Billion Yuan, Founder Says Company Is Chasing AGI, Not a Consumer App

DeepSeek has taken outside money for the first time. According to a circulated transcript reported by BigGo Finance, the Chinese AI developer closed a first external funding round exceeding 50 billion yuan, roughly $7.4 billion, with Tencent and CATL among the backers. This reverses the company's previous strategy of avoiding outside capital and steering clear of an IPO.

The transcript comes from a four-hour closed-door meeting between founder Liang Wenfeng and investors. IT Home first reported on the session, and the account was picked up by TechNode, BigGo Finance, and WEEX Crypto News, a crypto-focused outlet that credited RootData for the specifics.

Liang's message to investors: DeepSeek is not building a super app. He said the company will not chase 3D generation, video generation, or world modeling, and it will not close off its source code, according to WEEX. Products, in his framing, are "by-products" on the road to artificial general intelligence, not the point of the business.

Restraint as Strategy, Not Modesty

Liang repeatedly described "restraint" as DeepSeek's core competitive weapon, according to BigGo Finance. His logic: AI could eventually represent 10% of human GDP, a market so large that trying to monopolize it invites failure. "The more restrained you are, the more likely you are to succeed," Liang reportedly told investors. "Those who take more will be defeated by those who take less."

That philosophy extends to pricing. Liang said DeepSeek's API pricing isn't built around profit maximization, and that the company aims for "reasonable profits" rather than windfall gains, according to both BigGo Finance and WEEX. He reportedly added that DeepSeek is "still far from a complete shift to commercialization."

For a company that shook global AI markets with its low-cost R1 reasoning model, this admission carries weight. DeepSeek built its reputation partly on undercutting Western AI pricing. Liang's comments suggest that undercutting was never really about winning a pricing war. It was about buying time and market position while the company chases a much bigger technical goal.

The AGI Roadmap

Liang laid out a technical sequence: chain-of-thought reasoning, then AI agents, then continuous learning, according to BigGo Finance. He reportedly called continuous learning the next major technical challenge after agents, a point TechNode's coverage of the IT Home report also emphasized.

WEEX's account adds detail on sequencing: the current focus is coding agents and continuous learning capabilities, with the next phase aimed at autonomous model iteration, ultimately pointing toward embodied intelligence, meaning AI systems that operate in physical robots or real-world environments rather than staying confined to chat interfaces.

Open source remains central to that plan. Liang described open-sourcing DeepSeek's models as a long-term strategy to increase the probability of reaching AGI, not a marketing gimmick or a concession to competitive pressure, according to BigGo Finance.

China's Chip Gap, and Betting on a Breakthrough

On the U.S.-China AI competition, Liang reportedly said the real gap isn't talent, it's computing resources. He's optimistic China's domestic chip ecosystem will break through within a year, according to BigGo Finance, and claimed that with a fraction of the computing power available to U.S. labs, DeepSeek can close the capability gap to "6 months or 3 months," per WEEX.

That's an assertion from a founder pitching investors and defending his company's strategic position, not an independently verified benchmark. Nvidia's export-controlled chips and Huawei's domestic alternatives remain a real bottleneck for Chinese AI labs, and no source here provides hard evidence that China's chip ecosystem is a year from parity. It's Liang's stated belief, not a demonstrated fact.

On where the industry ends up, Liang predicted the market will converge around a small number of players, telling investors domestic large models would eventually settle into "two large companies and two small companies," according to WEEX. He named team stability as the one non-negotiable priority for DeepSeek internally, describing the company's culture as "a group of ordinary people accomplishing extraordinary things," per WEEX's account of the transcript.

What's Unverified

None of the three sources reviewed here are primary DeepSeek statements or an on-record company press release. All three trace back to a transcript that circulated after the meeting, originating with IT Home's reporting. No source names which specific funds beyond Tencent and CATL joined the round, and no source specifies DeepSeek's post-money valuation.

Whether restraint actually holds up as strategy once DeepSeek is sitting on $7.4 billion in fresh capital and answering to outside shareholders like Tencent remains an open question. Investors who just wrote large checks tend to eventually want commercial returns, reasonable or otherwise.