20 billion valuation, Alibaba and Tencent compete to invest, whose money will Liang Wenfeng take?

Null

Text | World Model Factory

DeepSeek financing reshuffle.

On April 22nd, Reuters, citing The Information, reported that Alibaba and Tencent are in talks to invest in DeepSeek, with valuation expectations rising above $20 billion.

Just a few days ago, Reuters reported that DeepSeek was seeking at least $300 million in funding, with a target valuation of no less than $10 billion.

In just a few days, DeepSeek’s valuation expectation has doubled.

As of press time, DeepSeek did not comment, and Reuters also stated that it had not independently verified all details.

But market signals are real.

Founded nearly two years ago, DeepSeek has always relied on its parent company, Fantom Quant, for funding, never accepting external investment.

Founder Liang Wenfeng is one of the most well-known independent figures in China’s AI circle—non-aligned, no fundraising, no going public.

Now, this crack has appeared.

When Alibaba and Tencent, two major giants, appear at the negotiation table simultaneously, what are they actually competing for?

Who will ultimately get DeepSeek’s funding?

DeepSeek’s market value

To answer what Alibaba and Tencent are fighting over, we first need to answer another question:

What does a $20 billion valuation really mean in China’s large model scene?

Considering the dark side of going public in Hong Kong, the latest round valuation is about $18 billion;

MiniMax’s Hong Kong IPO in January 2026 is estimated at about $6.5 billion;

Zhipu’s planned Hong Kong listing at the end of 2025 is estimated at about $6.7 billion.

If DeepSeek’s current round truly values it at $20 billion, then it’s basically the most expensive among China’s entrepreneurial large model companies.

Given the actual influence of its V3 and R1 models globally, this number is more like a starting point for negotiations rather than a ceiling.

As valuation rises, how much will DeepSeek raise?

In a public report on April 17th, DeepSeek was originally discussing raising at least $300 million at a valuation of at least $10 billion, roughly a 3% equity stake.

Using the same logic, at a $20 billion valuation, selling 3% equity would correspond to about $600 million; 5% would be about $1 billion; 10% would be about $2 billion.

So, how big is this funding for DeepSeek?

$300 million to $600 million is not a small amount for DeepSeek, but it’s not enough to be considered cash-rich.

Reuters emphasized in its report:

Now, cutting-edge models, inference, and agent systems are increasingly expensive, and DeepSeek’s initial external funding itself indicates it faces higher capital expenditure pressures.

These few hundred million dollars are more like opening an external financing channel, adding some ammunition, rather than securing enough funds for the next few years’ battles all at once.

If it can finally raise over $800 million to $1 billion, the situation changes.

This scale of funding is already very large among China’s entrepreneurial large model companies.

For comparison, MiniMax raised about $619 million in its Hong Kong IPO.

If DeepSeek’s single round of funding approaches or exceeds this level, it’s enough to show that the market regards it as a top-tier scarce asset among China’s entrepreneurial large models.

Regardless of the final range, this money itself is not the goal.

For Liang Wenfeng, two things matter most:

• Giving core employees a market-based option price to stop the crisis of talent loss;
• Securing non-financial resources such as compute discounts, cloud customer channels, supply chain collaboration, to support the rising capital demands in inference models and agent competition.

Alibaba and Tencent’s entry

No matter how high the valuation, it cannot stop continuous interest from investment institutions wanting to connect with Liang Wenfeng.

DeepSeek remains one of the most difficult projects in China’s AI investment circle to invest in.

An investor close to DeepSeek commented: This isn’t an asset you can buy just by offering the right price; in Liang Wenfeng’s screening criteria, money is the least important factor.

According to media reports, DeepSeek has previously rejected investment proposals from top Chinese venture capital firms and tech giants.

In the capital market’s view, DeepSeek has always been a scarce asset sought after. That’s why both Alibaba and Tencent want to squeeze into negotiations.

The answer isn’t that DeepSeek needs them, but that they need DeepSeek.

As China’s large models enter a phase of ecosystem, entry point, and resource battles, Alibaba and Tencent are unwilling to be absent from this key position at DeepSeek.

If they can invest, DeepSeek could become part of the ecosystem collaboration of major tech giants: model capabilities, cloud resources, enterprise clients, distribution interfaces—all have the potential to link up.

But if they can’t get in, or if others occupy the position first, DeepSeek might become a relatively independent yet sufficiently influential variable, potentially weakening the control of big firms in the AI era.

This type of investment is both offensive and defensive.

But deeper anxiety stems from the countdown of scarcity.

Zhipu and MiniMax are already listed; Moonlight Shadow is rushing toward a Hong Kong IPO; models like Leap Star and Baichuan Intelligence are accelerating capitalization.

DeepSeek is the only top general large model among the “Six Little Tigers” that has never accepted external funding.

Missing it means missing the last opportunity to get on board in this cycle.

Once it accepts other capital or goes public, Alibaba and Tencent will lose the chance to exchange equity for high-quality cloud clients.

For two giants with a combined market value exceeding $800 billion, a few hundred million dollars is just a small part of annual capital expenditure.

But if the opponent takes DeepSeek, what’s lost is the ecological dominance of an era.

Will DeepSeek take money from the big firms?

Alibaba and Tencent’s anxiety is real, but DeepSeek’s sense of urgency is equally undeniable.

Liang Wenfeng’s stance is clearly changing: from long-term rejection of external funds to opening a window for external financing for the first time.

Behind this may not only be a funding issue but also pressures from time, talent, and infrastructure starting to converge.

First, the pressure of model iteration—flagship model V4 has been delayed multiple times and has yet to debut;

Second, the talent defense—without external valuation, incentive options for core talent are significantly diminished;

Finally, infrastructure burden—adapting V4 to domestic chips requires massive engineering resources, which Fantom alone cannot support.

But Liang Wenfeng’s negotiation stance remains a crack open rather than a plea for entry.

The $20 billion valuation expectation, engaging with two giants, and delaying details all indicate he’s waiting for the most optimal solution.

This also means Liang Wenfeng cares not just about accepting or rejecting Alibaba or Tencent’s investment, but about which type of money and to what extent.

The first possibility is the most ideal and also the most scarce: only taking financial investment, without ceding strategic control.

In other words, big firms can buy into, but only as minority shareholders, without deeply binding business, demanding exclusivity, or locking DeepSeek into a specific ecosystem prematurely.

For DeepSeek, this is the most desirable scenario. The money is in, the relationship is established, but externally it’s still seen as a relatively independent model company.

The advantage of this structure is that it both supplies ammunition and preserves the most valuable independence in valuation.

The second, more pragmatic and riskier scenario: superficially taking financial investment, but actually binding resources.

Because big firms’ money rarely is just money; it usually comes with cloud resources, customer channels, interface capabilities, distribution access, and future cooperation expectations.

In the short term, this binding is tempting.

But the problem is, once the cooperation deepens, the market will quickly redefine DeepSeek—whether it remains an independent model company or becomes a key piece in a big firm’s AI landscape.

DeepSeek knows well what big firms’ money can bring: faster resource integration, larger commercialization potential, and a thicker safety cushion.

But it also knows that big firms’ money can erode external perceptions of its independence.

That’s the hardest part of this round of financing.

Once accepting investment, DeepSeek’s previously undefined state begins to change.

Who is the better choice?

If Alibaba and Tencent’s money come with ecosystem lock-in risks, is there a better option in the market?

Theoretically, state-owned or national team capital that offers resources but not tight control is closer to this balance.

State capital’s advantages are clear:

• Large funds, long cycles, no pursuit of short-term exit;
• Low intervention in technical routes, more tolerant of DeepSeek’s open-source strategy;
• Can also bring compute support, industrial collaboration, government project resources, and stronger policy coordination.

But the disadvantages are equally obvious:

• Long decision-making processes, high compliance requirements, possibly unable to keep pace with DeepSeek’s technical schedule;
• More subtly, the “state-owned” label might restrict international expansion, raising concerns among overseas developers and enterprise clients about data security and neutrality.

Currently, no clear reports indicate state funds are engaging with DeepSeek.

But Reuters reported in February 2025 that China Investment Corporation and the National Social Security Fund expressed interest in investing in DeepSeek.

This clue suggests that the national team is not without willingness.

And what about pure financial investors?

Top-tier institutions like Sequoia China and Hillhouse have money and professionalism, but their real shortcoming is resources.

DeepSeek isn’t short of money; it lacks the infrastructure and industry resources that come with it.

Industry insiders say that in recent months, many top VCs have tried to contact Liang Wenfeng, but most haven’t even gained the chance for formal due diligence.

Taking VC money, DeepSeek still faces commercialization demands, which fundamentally contradicts Liang Wenfeng’s original intention of “doing research” with DeepSeek.

Ultimately, this isn’t just a matter of choosing money; it’s about choosing identity.

Taking big firm money might quickly bring resources, accelerate commercialization, and help survive fierce competition, but at the cost of gradually eroding independence.

Taking state funds might preserve its technical route and open-source ideals but could slow down commercialization due to less efficient industry collaboration, ultimately becoming a strategic national asset.

Two types of money, two fates.

There’s no perfect answer—only choices based on trade-offs.

The window is counting down

With DeepSeek’s technological reputation and current market heat, it can always find buyers willing to pay.

But this seller’s position has a shelf life.

In the next 6 to 12 months, companies like Leap Star and Moonlight Shadow may gradually go public; the scarcity of top-tier unlisted general large models is rapidly diminishing.

Once these assets are all securitized, capital will shift its focus to secondary markets, and the premium space in private markets will shrink.

As more top global AI companies pursue financing or other capitalization moves, the valuation benchmarks for AI assets worldwide may continue to shift, and domestic valuation systems will face recalibration.

$20 billion today is an exciting starting point; tomorrow, it might be a price worth defending.

Liang Wenfeng’s crack has already appeared, but the door is only ajar.

How wide that door opens, and who gets to step through first, the countdown for that decision has already begun.