China’s DeepSeek AI Prepping for IPO, as It Becomes the Top AI Model

Listen to the audio version of this article (generated by AI).

China’s DeepSeek is planning to file for an initial public offering (“IPO”) as early as later this year, marking the emergence of a major Chinese AI model company into the public markets, according to Bloomberg. The late-year filing would allow the company to officially debut in early 2027.

The startup would be listed in mainland China, according to initial reports, though DeepSeek has only begun consulting banking advisors on the process.

The news comes as DeepSeek became the most used AI model company in June, amid a surge in the use of Chinese AI models. DeepSeek made waves among investors last year when it produced a high-powered AI model that could operate at a fraction of the cost of U.S. models.

At the same time, DeepSeek’s market valuation is also substantially lower than that of leading American AI model companies, OpenAI and Anthropic.

DeepSeek raised $7 billion in a June funding round, valuing it at about $50 billion, raising money from Tencent Holdings (TCEHY) and Contemporary Amperex Technology (CYATY), among others. It’s now looking to raise another round of funding at a $71 billion valuation, says Bloomberg, targeting at least another $1.5 billion in capital, though it could raise even more.

In contrast, Anthropic and OpenAI are valued much closer to $1 trillion – $965 billion and $852 billion, respectively – after multiple rounds of massive fundraising.

One of the biggest questions for investors is whether DeepSeek is currently profitable or at least has a path to profitability. The IPO filings will reveal the firm’s economics. American AI firms such as OpenAI are incurring significant losses that require steady cash infusions, and many investors are questioning these firms’ ability to ever turn a profit.

The question of DeepSeek’s profitability is perhaps even more pressing, given that its models run at substantially lower token costs. On this front, the news may not be particularly attractive to investors. Executives have said that DeepSeek will prioritize innovative AI research over short-term commercialization, hinting that profitability will be of secondary importance.

Whether American investors will get access to DeepSeek stock via a U.S. listing remains to be seen, though memory maker SK Hynix (SKHY) recently added a Nasdaq listing to complement its primary listing in South Korea. Regardless, it’s an ideal time for DeepSeek to consider an IPO.

Chinese AI Keeps Taking Share From American AI

DeepSeek’s IPO news comes as Chinese AI models continue taking market share from American AI models coded by OpenAI, Anthropic, and Alphabet (GOOGL), among others. 

In the last week of June, Chinese AI models processed 48% of tokens, compared to just 20% from American models, according to AI model brokerage platform OpenRouter. That’s a massive gain from last year, when American models led Chinese models by a 74% to 20% margin. 

DeepSeek models claimed about 16.3% of all traffic in June, setting it atop the leaderboard. Chinese open-source models claimed six of the top 10 spots for the month, including Xiaomi, Tencent, and MiniMax. Meanwhile, the American companies Anthropic, Google, and OpenAI held second, third, and fourth places, respectively. 

The key reason for the surge of Chinese AI models? They’re cheaper and almost as good. Many customers are finding out that they don’t need the highest-end frontier model from Anthropic for their business needs. A model that’s 90% as good – and totally competent for tasks such as drafting reports and emails, for example – at 10% of the cost works better for most users.

It’s worth noting that cost is a huge factor for power users. American models may cost 10-20 times more than Chinese models, and in some cases 150 times more. That’s why 80% of startups using open-source software are using Chinese AI models, according to Andreessen Horowitz.

With plenty of low-cost Chinese models and the uncertain return on investment of higher-priced models, companies are reconsidering whether they need top-of-the-line inference. Much cheaper and perfectly acceptable AI models can fill this gap quickly, and companies will pivot. 

So, given DeepSeek’s ascent to the top of the leaderboard, at least for June, and the attractive user economics of open-source models, it’s an ideal time for the DeepSeek news to leak.

DeepSeek IPO Comes Amid Frenzy for AI Companies to Go Public

DeepSeek is the latest name to join the IPO parade in a bid to capitalize on the intense investor interest in AI. The year 2026 is already shaping up to be one of the largest-ever years for IPOs, with AI-related companies dominating the largest offerings:

• Cerebras Systems (CBRS) conducted the largest-ever semiconductor IPO in May.
• SpaceX (SPCX) launched the largest IPO ever in June.
• Anthropic filed its IPO prospectus in early June, with some investors expecting an October offering valued at more than $1 trillion.
• Just days after Anthropic’s filing, OpenAI filed its own prospectus, though it quickly walked back the timeline, possibly to next year. It was last valued at $852 billion in March.

Major AI companies, and now DeepSeek, are looking to hit the IPO window while it’s wide open, even as serious concerns remain about whether major AI modelers like OpenAI and Anthropic can ever sustainably make money. OpenAI lost $20.9 billion in 2025 and an estimated $7 billion more in the first quarter of 2026, so it will be interesting to see whether DeepSeek is also operating in the red.

The revelation of DeepSeek’s financials and those of other AI modelers will be a key moment in helping investors decide whether AI firms have a reasonable path to profitability. With published financials, investors will be able to see the numbers for themselves and won’t have to accept AI hype. In turn, that will have a huge impact on the direction of AI investment and investors’ sentiment.

The rapid emergence of the highest-profile AI stocks should be a sign that AI mania is at a fever pitch, much like “dotcom” mania was in late 1999 and early 2000. Even if you believe AI is here to stay, that doesn’t necessarily make AI stocks a good buy – in the same way that Internet stocks weren’t a good investment back in 2000, even as the Internet went on to revolutionize business and the world at large.

Regards

James Royal, PhD

Editor’s Note: Marc Chaikin, the founder of Chaikin Analytics, has built an award-winning system that turned “bearish” on software stocks two months before they crashed this year. Now, he’s warning that one AI lab’s breakthrough could CRASH the Nasdaq while igniting a $500 trillion wealth transfer. This 60-year Wall Street legend has found a little-known $40 “pre-IPO backdoor” into the private startup behind this economic sea change. Click here for its name and ticker symbol.