Author: Omnitools
In May 2026, Moonshot AI was revealed to have reorganized its corporate structure back from B2B to B2C, with its head hospitalized due to high-intensity work and cost pressure. Almost simultaneously, the company completed a new funding round of approximately $2 billion, pushing its valuation past $20 billion. On one side is 'burning money until people can't take it anymore,' and on the other is capital queuing up to get in.
This is not an isolated case. Data from CCTV Finance shows nearly 600 funding rounds in China's AI sector in Q1 2026, with total funding exceeding 110 billion RMB, a sharp increase of 185.4% year-on-year. In May alone, Moonshot AI and StepFun, two large model companies, secured over 30 billion RMB. The embodied AI track was also hot, with over 50 disclosed funding rounds in Q1, accumulating about 20 billion RMB.
When single-quarter funding surpasses 100 billion RMB and the funding amounts for leading companies are dozens of times their annual revenue, what direction is this money pushing China's AI competition? OmniTools reviewed the domestic AI projects that received the most funding in Q1 2026 based on public financing data, attempting to answer three questions: Who is getting the money, what are they doing, and what do the trends behind this mean?
Where Did the Money Go? Nearly 600 Funding Rounds, Computing Power Consumes 30-50%
First, look at the macro picture. Citing data from venture capital institutions, CCTV Finance reported that the total funding in China's AI sector in Q1 2026 exceeded 110 billion RMB across nearly 600 funding events. What does 110 billion RMB represent? It's nearly triple the amount from the same period in 2025. Even in the history of global tech financing, this is a number worth noting.
The money mainly went to two tracks. Large model companies took the lion's share; in May alone, Moonshot AI and StepFun raised over 30 billion RMB. Embodied AI followed closely, with over 50 disclosed funding rounds in Q1, more than 30 companies receiving investment, totaling about 20 billion RMB, a year-on-year increase of nearly 60%.
Where the money is going is more noteworthy. The CCTV Finance report clearly pointed out three major directions: R&D, computing power, and talent. Leading large model companies' R&D expenditures in 2025 generally reached tens of billions of RMB; GPU procurement and cloud service leasing accounted for 30% to 50% of the funding amounts; and attracting top global talent was the third major expenditure.
The fact that 30% to 50% is being poured into computing power is key to understanding this funding wave. It means that for every 1 billion RMB in funding, 300 to 500 million RMB directly becomes hardware and cloud service bills. This is not wastefulness, but rather the cost of admission at the current technological stage. Training a generation of large models requires clusters of tens of thousands of GPUs, and inference services require widely distributed computing nodes. These are not problems that can be solved with a light asset model. The essence of this funding is the rapid conversion of capital into hardware barriers.
At the same time, the figure of tens of billions in R&D investment needs to be viewed against a backdrop: most large model companies' revenues in 2025 were far below this level. The model of annual revenue in the billions and R&D investment in the tens of billions means that financing is the only lifeline sustaining competition.
Large Model Track: Three $10 Billion+ Players, Three Different Playbooks
Among the three large model companies that attracted the most attention in this funding wave are Moonshot AI, StepFun, and DeepSeek. Their commonality is that their valuations have already exceeded or are approaching $10 billion, but their approaches are distinctly different.
Moonshot AI: Open-Source Route Accelerates Commercialization
Moonshot AI has had the densest funding rhythm in the past six months. According to reports from Securities Times and Investment Community, from December 2025 to May 2026, Moonshot AI completed multiple funding rounds within half a year, accumulating over $3.9 billion (approximately 37.6 billion RMB). Its valuation soared from $4.3 billion to $20 billion. In half a year, the valuation multiplied nearly fivefold.
Supporting this valuation is the commercial performance following the release of Kimi K2.5. Investment Community cited data showing that revenue in the 20 days after K2.5's release exceeded that of the entire year 2025, and its global payment ranking rose from outside the top 100 to 9th place. This means Moonshot AI's commercialization is not just 'showing signs of life' but has experienced a turning point-level leap.
The open-source strategy of K2.5 is the key variable driving this turning point. The model became the only Chinese open-source model integrated into Cursor and the official main model for OpenClaw, directly penetrating the overseas developer ecosystem. Silicon Valley VC a16z co-founder Marc Andreessen evaluated it as 'basically replicating GPT-5-level reasoning capabilities.' Investor Chamath Palihapitiya called it the 'Kimi K2.5 moment'—the first time an open-source model truly shook the closed-source system.
StepFun: Trillion-Parameter Foundation Model, Betting on Edge Deployment
StepFun is taking a different path. In January 2026, the company completed a B+ round of over 5 billion RMB, setting a then-record for the largest single round in the large model track in 12 months. In May, another new round of nearly $2.5 billion in funding was nearing completion.
Unlike Moonshot AI's open-source route, StepFun emphasizes the parameter scale of its foundation model and edge deployment. Its trillion-parameter Step-2 model is positioned as an enterprise-grade foundation, with multimodal capabilities covering scenarios like visual understanding and video generation. On the product side, Step 3.5 Flash topped the OpenRouter fastest model chart on its release day.
More noteworthy is who is investing in StepFun. In the May funding round, smartphone industry chain companies like Huaqin, Longcheer, Omnivision, and ZTE collectively entered. This is no coincidence. Yin Qi, former co-founder of Megvii, has taken the role of Chairman at StepFun. The company has dismantled its VIE structure and is sprinting towards a Hong Kong IPO. The entry of smartphone industry chain capital means StepFun's strategy is not to be a pure software company, but to embed AI capabilities into edge devices, pursuing a 'model + hardware' integrated route.
DeepSeek: From No Funding to a $45 Billion Valuation
DeepSeek's first funding round was one of the most symbolic events of 2026. According to reports from Securities Times and Caijing, DeepSeek has initiated its first funding round, with its valuation soaring to $45 billion. It plans to raise over 50 billion RMB, with the Big Fund in talks to lead the investment.
Prior to this, DeepSeek had never conducted external financing. Founder Liang Wenfeng personally invested 20 billion RMB to support the company's operations. DeepSeek is known for its open-source models and extreme cost-effectiveness. Its technical approach has exerted significant downward pressure on API prices across the industry, pushing mainstream commercial models into the 'cent era'.
'Even companies that don't need money are starting to take it'—this signal is more important than any funding figure. It indicates that the intensity of the current AI competition has reached a critical point: even with technological and cost advantages, facing the industry-wide computing power arms race and talent war, not raising funds could lead to falling behind. The Big Fund's negotiations to lead the investment mean state capital is formally entering the core layer of large models. The $45 billion valuation corresponds to a company that has not yet scaled its commercialization, and the pressure to justify this valuation is self-evident.
Embodied AI: Over 10 Companies Valued at 10B+ RMB: Boom or Bubble?
Embodied AI was another main line of funding in Q1. Data from IT Juzi and Securities Times show over 50 disclosed funding rounds in this track in Q1, accumulating about 20 billion RMB, a year-on-year increase of nearly 60%. Statistics from Guangzhou Daily show over 10 companies have already broken the 10 billion RMB valuation mark.
Galaxy General is the highest-valued project in this track. In March 2026, the company completed a 2.5 billion RMB funding round, with a valuation exceeding 20 billion RMB. Investors included Big Fund III, Sinopec, and CITIC. This was Big Fund III's first investment in the embodied AI field. Galaxy General's core products are general-purpose humanoid robots and the 'Galaxy Star Brain' embodied large model system. Its robots cooperate with CATL, achieving 7×24-hour fully autonomous operation in battery factories.
Qianxun Intelligence completed two consecutive rounds totaling nearly 2 billion RMB, reaching a valuation of over 10 billion RMB in 26 months since its founding, with participation from Yunfeng Capital, Chaos Investment, Sequoia Capital, etc. Qianxun's humanoid embodied intelligent production line has been deployed at CATL's Zhongzhou base.
Independent Variable Robotics completed a 1 billion RMB A++ round and a several hundred million RMB strategic round, with a valuation exceeding 10 billion RMB. It is the only embodied AI company simultaneously invested in by all three of China's internet giants: Alibaba, Meituan, and ByteDance. Founder Wang Qian is one of the earliest scholars globally to introduce the attention mechanism into neural networks. Independent Variable pursues a full-stack in-house R&D route for both software and hardware and has launched the 'Quantum One' and 'Quantum Two' humanoid robot bodies.
Zhi Jian Power is the most special project in this funding wave. The company was founded only in July 2025 and raised five rounds in 8 months, accumulating 2 billion RMB, with a valuation exceeding $1 billion, becoming the youngest embodied AI unicorn. Sequoia, Tencent, and Alibaba all participated in the investment. Achieving a valuation over $1 billion in less than a year is extremely rare in the hardware track.
Lingxin Qiaoshou focuses on a single category, achieving perfection. The company completed a nearly 1.5 billion RMB Series B round, with a valuation over 10 billion RMB. It is the world's only manufacturer producing a thousand high-degree-of-freedom dexterous hands per month, holding over 80% of the global market share. By becoming a global leader in a niche hardware category, Lingxin Qiaoshou's path differs from humanoid robot companies making complete machines; it's more like a 'hidden champion' in the embodied AI industry chain.
Vita Power and Luming Robot represent two directions: consumer-grade and industrial capital. Vita Power's Pre-A round raised nearly 500 million RMB, a record for the largest single round in the consumer-grade embodied AI track. The Vbot super-powered robotic dog has started initial deliveries of 500 units. Luming Robot accumulated nearly 1 billion RMB in funding, led by Mitsubishi Electric. The founder is a former senior executive from Dreame. Revenue in 2026 is expected to reach the hundred-million RMB level.
The embodied AI track is mass-producing 10-billion-RMB unicorns; that is a fact. However, very few companies have disclosed revenue. Unitree Technology's 2025 revenue was 1.708 billion RMB; Leju Robotics' revenue was 258 million RMB and applied for an IPO—leading companies indeed have real income. But most 10-billion-RMB valuation companies are still in the early validation stage. The chasm between valuation and revenue is the biggest uncertainty. When over ten companies simultaneously break the 10-billion-RMB valuation mark, the pressure of an industry valuation bubble cannot be ignored. The high expectations of capital ultimately need to be realized on client sites and production lines.
Behind the Ledger: Five Emerging Trends
Based on the above funding data and project review, OmniTools summarizes five emerging trends.
Trend 1: Capital structure is shifting from VC-led to a mix of industrial capital and state-backed funds. Big Fund III investing in Galaxy General, Big Fund negotiating to lead DeepSeek's first round, Mitsubishi Electric leading Luming Robot's round, smartphone industry chain capital investing in StepFun. Industrial capital brings orders and application scenarios; state-backed funds bring strategic intent. The proportion of purely financial investors is decreasing. For companies, this means that taking money is not just about delivering returns but also about being responsible for industrial implementation and compliance.
Trend 2: Valuation ceilings are rapidly rising, and the Matthew Effect is intensifying. Moonshot AI at $20 billion, DeepSeek at $45 billion, Galaxy General at 21 billion RMB—top companies are absorbing the majority of funds. Meanwhile, in the embodied AI track, over 30 companies received funding, but only about 10 raised over 1 billion RMB. As funds concentrate at the top, the risk of consolidation for mid-to-lower tier projects is rising.
Trend 3: IPO channels are accelerating. Leju Robotics' ChiNext IPO has been accepted, planning to raise 2.6 billion RMB, becoming the first AI company to apply the ChiNext's fourth set of listing standards. StepFun dismantled its VIE structure and is sprinting for a Hong Kong listing. Zhipu and MiniMax have also initiated listing processes. The high valuations in the primary market need the secondary market to take over; 2026 to 2027 may become a concentrated listing window for Chinese AI companies.
Trend 4: Open-source routes are gaining differentiated competitive advantages. Kimi K2.5 penetrated the overseas developer ecosystem through open source. DeepSeek's open-source strategy lowered prices across the industry. Open source is no longer a 'free strategy' but a means to gain global developer entry and build ecosystem barriers. For downstream clients, this means more cost-effective model options and continuously declining API costs.
Trend 5: Embodied AI is moving from proof-of-concept to small-batch delivery. Vita Power's Vbot delivering its first 500 units, Galaxy General's robots operating at CATL, Unitree shipping 5,500 humanoid robots in 2025. It's not all PowerPoint anymore; some products are already working at client sites. However, the journey from 'small-batch' to 'large-scale' involves challenges in supply chain, quality control, after-sales service, and cost control—each stage is difficult.
The 110 billion RMB in Q1 2026 funding is pushing China's AI competition into a new stage. The characteristic of this stage is no longer 'whose model has more parameters,' but 'whose money can last until the day commercialization is realized.' R&D investments in the tens of billions, computing power consuming 30-50%, the batch emergence of 10-billion-RMB valuations—these figures are proof of the industry's vitality but also signals of a high-intensity war of attrition. When the pace of fundraising becomes a core competitive advantage, and GPU reserves determine model iteration speed, the final outcome of this wave of AI entrepreneurship may not be decided by a single technological breakthrough, but by cash flow and management capabilities—who can stay at the table.
