DeepSeek, the Chinese AI lab that upended the AI industry earlier this year with open-source models that matched closed proprietary competitors at a fraction of the cost, is closing in on its first-ever external funding round. Sources familiar with the deal say the company is finalising a raise of around $7.4 billion, with Tencent, battery giant CATL, and China’s National AI Industry Investment Fund among the investors expected to participate.

The round would value DeepSeek at between $52 billion and $59 billion — a figure that has risen sharply from the $20 billion valuation discussed in earlier talks just six weeks ago.

The Numbers Behind the Deal

The structure of the funding round is notable. Founder Liang Wenfeng is expected to contribute $3 billion of his own capital, signalling strong personal conviction in the company’s direction. Tencent Holdings is in line to invest approximately $1.5 billion, while CATL — the world’s largest EV battery manufacturer — is committing around $740 million.

Fewer than ten investors are expected to participate in total. The deal is expected to close within the next two to three weeks.

This marks a significant shift for DeepSeek, which has operated with unusual restraint for a company of its stature. For most of its existence, it was funded entirely by High-Flyer, the quantitative hedge fund that spun out the AI lab. Accepting outside capital at this scale is a strategic move, not a necessity.

Why This Is Bigger Than the Headline Number

The more interesting story is what this round signals about the direction of open-source AI.

DeepSeek’s models — particularly the R1 reasoning model and the DeepSeek V3 series — achieved something that most industry observers said was impossible: frontier-level performance from a team that spent a fraction of what American labs have spent on training compute. When those results landed in early 2026, they triggered a genuine rethink among enterprise technology teams about whether they needed to pay premium prices for proprietary AI access.

Now, with $7.4 billion in fresh capital, DeepSeek has the resources to continue that trajectory at a much larger scale. The expectation is that the money will go toward compute infrastructure, model training, and expanding open-source model releases — which means better and more capable models available to the broader development community.

For companies building AI-powered products and workflows, that has direct implications. Stronger open-source foundations mean lower costs, more flexibility, and less lock-in to any single vendor’s pricing and terms.

The Open vs. Closed AI Debate Gets a Data Point

The enterprise AI world has been slowly splitting into two camps.

On one side are closed-model providers: OpenAI, Anthropic, and Google. These companies charge for API access and continue to release progressively more capable models at increasingly premium price points. On the other side are open-source alternatives: Meta’s Llama family, Mistral, and DeepSeek. These models can be run on-premises or hosted independently, giving businesses more control but requiring more technical capability to operate.

Until recently, the argument for closed models was straightforward: they were simply more capable. That gap has narrowed considerably. And with DeepSeek securing the capital to accelerate, the gap may narrow further still.

The question for business leaders is no longer “open source or closed?” as a matter of principle. It is now a genuine trade-off between customisation and control on one hand, and convenience and guaranteed capability on the other.

The Geopolitical Dimension

It would be naive to read this as a purely commercial story. DeepSeek’s backers include China’s National AI Industry Investment Fund — a government vehicle explicitly designed to build national AI capability. CATL’s involvement is similarly strategic: the company has ambitions to use AI to accelerate its own battery chemistry research and manufacturing.

This shapes the context in which the round is happening. China is investing heavily to remain competitive in AI model development, and DeepSeek is one of its clearest success stories. For enterprise teams assessing their AI supply chain and vendor risk, that is a relevant consideration — particularly for organisations in regulated industries or those handling sensitive data.

The US government’s recent executive order on frontier AI models, which asks developers to voluntarily submit models for 30-day pre-release review, does not apply to foreign labs. That regulatory asymmetry is something procurement and legal teams at larger enterprises will need to think through.

What This Means for Business

If you are already using DeepSeek models: This round is broadly positive. More capital means more investment in model quality, tooling, and infrastructure. The open-source commitment is likely to continue.

If you are evaluating open-source alternatives to proprietary AI: DeepSeek’s funding validates that open-source AI is a sustainable path, not a corner-cutting shortcut. The cost advantages that made the models attractive have not gone away — and performance is improving.

If you are locked into a single closed-model provider: Now is a reasonable time to build your AI architecture in a way that allows you to switch. The gap between proprietary and open-source model capability is smaller than it has ever been, and the cost difference is still significant.

If you are watching the China-US AI dynamic: The funding round is a signal that the competition between American and Chinese AI development is not slowing down. Business leaders who assumed one side would pull ahead definitively should probably revisit that assumption.

---

Enterprise DNA works with businesses navigating AI adoption decisions — from model selection through to full workflow deployment. If you are trying to figure out where open-source fits in your AI strategy, talk to us about an advisory session.