The AI sector is abuzz with a significant development. According to media reports citing Bloomberg, DeepSeek is advancing a funding round targeting approximately 70 billion yuan (about $10 billion USD). Potential investors reportedly include national-level industrial funds, Tencent, and IDG Capital.
While the details have not been officially confirmed by DeepSeek and both the amount and investor list may change, the mere rumor has already stirred the market. This potential funding round represents a major shift for a company historically known for its reluctance to seek external capital and its focus on technical prowess over fundraising narratives.
The narrative around this funding has escalated rapidly. Initial reports in early May suggested DeepSeek was planning its first external financing, targeting a valuation over $10 billion and aiming to raise at least $3 billion. The discussion then centered on whether founder Liang Wenfeng had finally relented to external investment, given DeepSeek's unique position as a well-funded AI player backed by quantitative investment firm, 幻方量化 (Huanfang Quant).
By mid-May, the story intensified. Reports emerged that national industrial funds and other official investment institutions were in talks to lead the round, potentially valuing DeepSeek at around $45 billion. Subsequently, figures of a $50 billion valuation and a 50 billion yuan fundraising target surfaced. The latest market whispers now point to a staggering 70 billion yuan funding round.
Within just over a month, DeepSeek has transformed from a company that "might raise funds" to one potentially securing one of the most closely watched financing rounds in China's tech startup history. This evolution is what truly captivates the market. It is crucial to note that the 70 billion yuan figure refers to the intended fundraising amount, not the company's valuation, which some market sources suggest is being discussed in the tens of billions of dollars range. Regardless of the final numbers, one thing is clear: the capital markets are reassessing DeepSeek's worth.
Historically, DeepSeek's value stemmed from its technical breakthroughs. The debut of its R1 model was a revelation, demonstrating that a Chinese AI firm could compete at the global forefront, challenging the narrative of merely following OpenAI. DeepSeek impressed the global AI community with its cost-effective and open approach. The company's narrative was compelling—a Chinese team breaking new ground against well-funded US giants, prioritizing efficiency over brute-force computing power, and championing open-source ideals. This story captured the public's imagination.
However, capital markets are now looking beyond a single successful model. A breakthrough can propel a company into the spotlight, but it doesn't guarantee a permanent seat at the table. The AI industry's harsh reality is that today's star is immediately asked about its next model, computing resources, team stability, commercialization path, and ecosystem development. DeepSeek has proven it can win a stunning surprise attack. The current question is whether it can sustain a prolonged war. The escalating funding rumors essentially represent the market betting on this very capability. DeepSeek is no longer being priced as just a promising Chinese AI dark horse; it is increasingly being valued as a potential foundational AI infrastructure company for China.
Interpreting this funding round solely as Liang Wenfeng needing money underestimates the situation. While capital is necessary, what Liang and DeepSeek likely need more are time, top-tier talent, control, and a more substantial entry ticket into the next phase of AI competition. The financial demands of the current AI race are beyond the reach of ordinary entrepreneurs. Every incremental advance in a top-tier model consumes vast amounts of computing power, data, engineering talent, and R&D time.
DeepSeek's past strength was its efficiency, challenging the industry's assumption that massive compute spending was the only path to a powerful model. This story was inspiring and boosted morale for domestic AI. However, efficiency is not a permanent shield. Smarter methods can reduce costs, but they don't eliminate the need for investment. As competitors like OpenAI, Anthropic, Google, Meta, Alibaba, ByteDance, Tencent, and Baidu continue to pour resources and frequently update their models, the pressure on DeepSeek extends beyond simply building a good model. It must continue training more powerful models, ensure reliable inference services, adapt to domestic chips, build a developer ecosystem, serve enterprise clients, and retain the best talent—all long-term, resource-intensive endeavors.
A critical reality is the high cost of AI talent. DeepSeek previously attracted top young talent with its technical vision, research freedom, and global reputation. However, people don't live on papers and leaderboards alone. Large tech companies can offer substantial cash compensation, stock options, diverse business scenarios, and more predictable wealth creation. Without an external valuation benchmark, the value of employee stock options at DeepSeek remains unclear, which is a significant risk for a talent-intensive AI company. While public discourse often focuses on model parameters and training costs, the stability and motivation of the core team often determine a company's long-term trajectory. Therefore, the primary significance of this funding round for DeepSeek might not be to replenish its treasury, but to establish a valuation for its team. Once external investors are onboard, the company's valuation gains a market reference point, providing clearer benchmarks for employee incentives—a factor potentially more critical than a cash infusion for a company like DeepSeek.
Regarding control, business registration information shows that on April 27, 2026, DeepSeek's registered capital increased from 10 million yuan to 15 million yuan, and Liang Wenfeng's direct shareholding rose from 1% to 34%. After adjustments, he controls approximately 84.29% of the company's equity directly and indirectly. This move is significant. If DeepSeek is indeed preparing to welcome external capital, Liang's first step was to solidify his grip on the steering wheel. Capital brings resources but also pressure. Investors, even patient industrial or state-backed capital, will expect DeepSeek to assume a clearer industrial role. Liang understands this, which is why he isn't merely presenting a pitch deck to seek funds but has proactively restructured the company's control before opening the door to investors. This is a key aspect of this funding round: Liang Wenfeng isn't suddenly transforming from a technical idealist into a capital player; rather, reality is compelling him to shape the company into a more mature entity. The past DeepSeek operated like a special forces unit—agile and sharp. The future DeepSeek must evolve into a regular army, which relies on a robust system.
DeepSeek's past rise resembled a genius narrative: a low-profile founder, a team disinterested in fundraising, a group of young engineers suddenly unveiling an industry-shaking model just as global AI giants were at their peak. This story was immensely appealing, fulfilling many people's hopes for domestic AI—that Chinese companies could compete on core technology without relying solely on massive funding, aggressive marketing, or a charismatic founder.
However, the genius narrative has a flaw: it cannot be sustained indefinitely. The impact of R1 came from its surprise factor. The industry didn't anticipate DeepSeek reaching that level, so its debut was explosive. Now, DeepSeek stands firmly in the spotlight. Every update is compared against OpenAI, Anthropic, Google, Alibaba, and ByteDance. The outside world is no longer satisfied with mere strength; it demands sustained excellence. This is the true pressure of being a top-tier company.
Subsequent releases, like the V4 preview, which emphasized improvements in knowledge, reasoning, and agent capabilities, and signaled adaptation to the domestic chip ecosystem, failed to replicate the sensation of R1. This isn't because DeepSeek regressed, but because audience expectations have been raised. When a company transitions from dark horse to industry leader, surprises diminish, and responsibilities multiply. This evolution is precisely why DeepSeek must embrace capitalization.
The next phase of competition for China's AI industry is no longer just about topping model leaderboards. It encompasses battles for ecosystem dominance, computing resources, application scenarios, and top talent. If DeepSeek continues solely as a cool open-source model company, it retains significant value. But if it aims to become a central pillar of China's AI landscape, it cannot remain content with applause from the technical community. It requires more stable computing power, a larger talent pool, mature enterprise service capabilities, and deeper integration with domestic chips, cloud platforms, application ecosystems, and government/enterprise clients.
Put bluntly, DeepSeek is transitioning from being a hit model to becoming an AI infrastructure provider. The cost of this transition is evident. The most captivating aspect of the old DeepSeek was its departure from the typical AI company mold—it possessed an exceptional quality, unhurried by fundraising, monetization, or crafting a perfect market narrative. Post-funding, this sense of exception will inevitably diminish. The outside world will scrutinize its valuation, investors will monitor returns, the industry will watch its commercialization efforts, developers will question its commitment to open-source, and competitors will eye its talent stability.
DeepSeek will become stronger, but also heavier. This is the very dilemma Liang Wenfeng has long grappled with. In the past, Liang proved that excellent models could be built with less money. Next, he must prove whether a market-driven approach can enable DeepSeek to go the distance. This challenge is more difficult than creating a hit model. A model's strength is quickly validated by benchmarks and users, but a company's long-term strength requires time to prove.
The real question DeepSeek faces is not how staggering a 70 billion yuan funding round is, but what it will become after securing this ticket. This question is larger than 70 billion yuan. Liang Wenfeng's spring may have truly arrived, but it is not a light, carefree season. It marks the beginning of DeepSeek's journey from myth to reality, from a hit model to an infrastructure company. If it can withstand this transformation, China's AI sector will have gained a true long-term contender. If it cannot, then the 70 billion yuan will merely become another compelling capital story.