DeepSeek seeks $300M in first fundraise at $10B+ valuation as AI costs surge

DeepSeek is preparing to raise outside capital for the first time, a move that could value the Chinese AI startup at more than $10 billion as the cost of building advanced models continues to climb.

The company, which shook Silicon Valley and Wall Street last year with its low-cost R1 reasoning model, is in talks to secure at least $300 million, according to people familiar with the discussions. The move marks a clear shift for a company that has spent its short life avoiding external funding and focusing on research over growth at any cost.

“DeepSeek is in talks to raise outside capital for the first time, seeking to beef up its financial war chest so it can better compete in the costly battle to develop leading AI models,” The Information reported, citing four people familiar with the matter.

Until now, DeepSeek has been backed entirely by High-Flyer Capital Management, the quantitative hedge fund that also gave birth to the startup. That backing helped DeepSeek operate on its own terms, free from investor pressure and the usual expectations that come with venture capital.

That independence is starting to meet reality.

Demand for DeepSeek’s models and API services has surged, pushing its infrastructure to the edge. The company needs more GPUs, more servers, and far more compute capacity to keep up. Training and running advanced reasoning systems and emerging agent-style AI is expensive, even for a team known for doing more with less.

That tension—between efficiency and scale—is now driving DeepSeek toward outside capital.

The company’s rise has been hard to ignore. When DeepSeek released its R1 model in early 2025, it quickly drew comparisons to OpenAI’s top systems. What stood out wasn’t just performance, but the cost. R1 was reportedly trained for around $5.6 to $6 million using older, export-restricted Nvidia H800 chips, a fraction of what many U.S. labs spend.

That single data point shook assumptions across the AI market.

DeepSeek Moves to Raise First Outside Capital at $10B+ After Shaking AI Markets

Investors began questioning whether massive compute budgets were the only path forward. Chip stocks slid. The broader AI trade lost momentum. One estimate pegged the market reaction at roughly $1 trillion in erased value across AI-related companies. For a startup barely two years old, it was a rare moment of influence.

DeepSeek leaned into a different philosophy. Its models are open-source, and its inference costs are dramatically lower in certain use cases. Engineers focused on efficiency techniques such as KV cache compression and selective activation, proving that constraints can lead to smarter designs.

Even so, scale still matters.

The surge in usage has exposed limits. Outages have become more frequent. Infrastructure costs are climbing. The company’s internal funding, supported by High-Flyer’s strong performance, has carried it this far. Reports suggest the hedge fund posted returns of about 56.6% in 2025, giving DeepSeek a solid financial base. That cushion now looks smaller against the demands of global competition.

Founder and CEO Liang Wenfeng has long resisted outside investors, worried that external pressure could dilute the company’s research culture or complicate its global position. That stance helped shape DeepSeek’s identity. It built a reputation as a lab-first organization, not a startup chasing quick commercialization.

This fundraising effort signals a recalibration, not a complete shift.

The valuation itself reflects the uncertainty around DeepSeek. Estimates have ranged widely, from as low as $1 billion to well above $20 billion. A $10 billion-plus figure lands somewhere in the middle, capturing both its outsized impact and its relatively lean structure.

Who invests may matter as much as how much is raised.

Domestic Chinese investors are the most likely participants. U.S. venture firms face regulatory pressure and national security concerns that could limit involvement. Export controls on advanced chips remain a key factor shaping the company’s trajectory. DeepSeek has managed to work within those limits, mixing restricted hardware with growing use of domestic alternatives.

That balancing act sits at the center of a larger story.

The global AI race is no longer just about building bigger models. It is about who can build them efficiently, who controls the infrastructure, and who can sustain the cost over time. DeepSeek’s approach has already forced a rethink among its competitors. Companies like Anthropic and Google continue to push scale, yet the pressure to justify spending is rising.

DeepSeek’s next move could shape how that debate unfolds.

If the funding round closes, it would place the company among the most valuable AI startups in China and give it the resources to expand its infrastructure, stabilize its services, and advance commercialization. At the same time, it will test whether a research-driven culture can hold its ground once outside investors come into the picture.

For now, the startup that proved AI breakthroughs don’t require billion-dollar budgets is stepping into a new phase—one where efficiency alone may not be enough to win.