DeepSeek eyes $71B valuation one month after first raise

Key Takeaways

• DeepSeek is in preliminary talks for a new round at $71 billion valuation, up 37% from its $52B first round
• The company raised $7 billion in its debut financing, closed around late May 2025
• DeepSeek is developing its own AI chip to reduce dependence on Nvidia and Huawei

DeepSeek, the Chinese AI startup that rattled Silicon Valley with its efficient models earlier this year, is already shopping for more capital. The company has begun preliminary talks with new investors about a round that would value it at roughly $71 billion, the Financial Times reported Tuesday. That's a 37% jump from the $52 billion valuation it achieved just one month ago.

DeepSeek closed its first-ever funding round around the end of May, pulling in approximately $7 billion. The speed of this second approach is unusual. Most startups wait quarters, if not years, between major raises. DeepSeek did not respond to a request for comment.

Why the rush for more capital?

The company became China's national AI champion and grabbed global attention in early 2025 when its V3 and R1 models drew praise from researchers and engineers in the US. Those models challenged American assumptions about what Chinese labs could achieve, particularly on constrained compute budgets. DeepSeek reportedly trained R1 for around $5.6 million, a fraction of what OpenAI or Anthropic spend on comparable frontier models.

That efficiency came partly from algorithmic innovation and partly from necessity. US export controls limit China's access to the most advanced Nvidia GPUs. DeepSeek has relied on a mix of Nvidia and Huawei chips to train and run its models. But reliance on any external supplier creates risk, which brings us to the likely reason for the fresh funding.

DeepSeek is building its own AI chip

Reuters reported earlier this month that DeepSeek is developing its own AI chip. Custom silicon is expensive. Nvidia spent years and billions of dollars building its CUDA ecosystem. For DeepSeek to design, tape out, and manufacture a chip capable of training frontier models, it will need deep pockets and a long runway.

A $71 billion valuation, if achieved, would place DeepSeek among the most valuable private AI companies in the world, alongside OpenAI. The difference: OpenAI has raised over $20 billion across multiple rounds since 2019. DeepSeek would be approaching similar territory in a matter of months.

What this signals for US and Chinese AI competition

DeepSeek's rapid capital accumulation suggests Chinese investors see a window. Export controls have not stopped progress. In fact, they may have accelerated Chinese focus on efficiency and vertical integration. If DeepSeek can produce competitive models with less compute, and then build its own chips to remove the Nvidia dependency entirely, the US leverage from export bans diminishes.

Marc Andreessen called DeepSeek "a wake-up call for the US AI industry" when the R1 model launched. He noted they achieved comparable results with a fraction of the compute resources. That wake-up call is now raising capital at a pace that matches the alarm it triggered.

The founder's background

DeepSeek was founded in 2023 by Liang Wenfeng, who also co-founded High-Flyer, a quantitative hedge fund. That finance background shows in the company's approach: maximize returns per compute dollar. The hedge fund connection also provided early access to GPUs and capital before the first formal round.

Frequently Asked Questions

How much did DeepSeek raise in its first funding round?

DeepSeek raised approximately $7 billion in its first-ever funding round, which closed around the end of May 2025, at a valuation of $52 billion including the raised funds.

What valuation is DeepSeek targeting in its new round?

DeepSeek is in preliminary talks with investors about a round that would value the company at approximately $71 billion, representing a 37% increase from its previous valuation.

Why is DeepSeek developing its own AI chip?

DeepSeek is building custom AI silicon to reduce its reliance on Nvidia and Huawei chips. US export controls limit China's access to advanced Nvidia GPUs, making vertical integration a strategic priority.

What models made DeepSeek famous?

DeepSeek's V3 and R1 models gained widespread praise in Silicon Valley in early 2025. They challenged assumptions about China's AI capabilities by achieving competitive performance at a fraction of typical training costs.

Source: Tech-Economic Times / ET