Summary of Key Points
In the first half of 2026, domestic AI financing in China has truly exploded: the total funding for the half-year exceeded 300 billion yuan, far surpassing the entire amount of 2025. Beijing, Hangzhou, Shanghai, and Shenzhen were the core cities for financing, with Hangzhou rising to second place due to DeepSeek’s $510 million in funding. The large-model segment accounted for more than half of the total funds, with embodied intelligence (humanoid robots, etc.) being the most active sub-field. Capital prefers companies in both growth and mature stages, with early investments shifting towards downstream applications of large models such as “world models.” The financing volume for the second half of the year is expected to exceed 600 billion yuan, but the large-model sector will face a period of consolidation.
I. Explosive Financing Growth in the First Half of the Year
From January to June 2026, there were 1,203 financing events in China’s AI sector, with a total amount exceeding 300 billion yuan—more than the entire year of 2025. The pace of financing was rapid, with monthly amounts consistently above 40 billion yuan from March to May, and in June, it soared to 100 billion yuan, mainly due to DeepSeek’s $510 million in Series A funding (equivalent to the market value of a mid-sized listed company).
Geographically, the four cities of Beijing, Shanghai, Hangzhou, and Shenzhen accounted for 74% of the financing events and 86% of the funds: Beijing led with 321 events and 955 billion yuan; Hangzhou overtook Shanghai and Shenzhen thanks to DeepSeek’s funding and became the new leader among first-tier cities. Shanghai (595 billion yuan) had a stronger fundraising capacity than Shenzhen (359 billion yuan), indicating higher overall valuations of AI companies there. Suzhou, with nearly 100 million yuan in funding from Momenta, demonstrated a significant clustering effect in the Yangtze River Delta region. Additionally, new first-tier cities like Hefei and Chengdu also saw AI-related financings, with some even achieving their first investments.
II. Where Did the Money Go? Large Models Dominate, Embodied Intelligence Was the Hotspot
The funds in the AI sector were not distributed evenly but concentrated in several key areas:
- Large models dominated with 159.8 billion yuan (over 50% of the total), with leading projects securing hundreds of millions in funding, such as DeepSeek’s $510 million. This suggests that capital views large models as the foundational technology for AI and is willing to invest heavily in them.
- Embodied intelligence (humanoid robots, etc.) was the most active with 906 billion yuan in financing and 312 events (26% of the total), making it the most project-heavy area. For example, Zixiang Robot received 6.3 billion yuan in just half a year, reflecting capital’s bet on who will be the first to mass-produce these technologies.
- AIGC applications were the most mature with 596 billion yuan in funding, representing the best stage for commercialization (e.g., video/image generation). However, no single company has yet dominated this field, indicating ongoing competition for market share.
- Fundamental layers (computing power, chips) were also crucial with 725 billion yuan in financing; these are capital-intensive areas that support both large models and embodied intelligence.
In terms of individual project funding, AIGC (average of 710 million yuan) and large models (704 million yuan) had the highest amounts, indicating that capital is willing to offer high premiums to leading companies in these fields.
III. How Does Capital Invest? Growth Stage is the Main Battlefield; Early Investments Focus on “World Models”
The logic behind capital investment in AI is clear:
- Early stages (seed/angel rounds): A broad approach with a focus on the future, accounting for half of the events (626), but with an average investment of only 73 million yuan per deal, reflecting low-risk experimentation. Currently, “world models” (the “operating systems” for embodied intelligence that enable robots to understand the physical world) are in high demand, with six companies receiving 9.7 billion yuan each. Angel round funding has expanded—while previous rounds typically amounted to a few hundred million yuan, companies like Bragg and Wujiedong received 2.1 billion and 1.9 billion yuan respectively due to the scarcity of top-tier teams. Additionally, talent from large companies starting startups (e.g., the Daxiao Robot incubated by SenseTime) is becoming a trend, indicating that the early window for pure large-model startups has closed.
- Growth stage (Series A/B rounds): The core battlefield, accounting for 49.4% of total funding, with an average investment of 380 million yuan per deal. Companies at this stage have passed technical validation and have product and commercialization potential, such as Jieyue Xingchen, which completed four rounds of financing in half a year and is preparing for listing.
- Mature stage (leaders): These companies are considered the “stabilizers” of the industry, with only 177 events but over 100 billion yuan in funding, averaging 624 million yuan per deal. Examples include Kimi, which received 18.9 billion yuan. Capital invests in them for stable returns.
Among the top 20 companies in the later stages, the three large-model giants (DeepSeek, Jieyue Xingchen, Kimi) accounted for 60% of the funds, with seven humanoid robot companies accounting for 18%, and the AIGC sector’s three leading players accounting for 4%. Other sectors included autonomous driving (Momenta) and AI pharmaceuticals (Huashen Zhiyao).
IV. Outlook for the Second Half of the Year: Total Financing Could Exceed 600 Billion Yuan, but Large Models Face Consolidation
Based on the first-half trend, annual financing is likely to exceed 600 billion yuan. However, the pace will slow down in the second half due to the high base from the first half. The most significant change is that the large-model sector will begin to consolidate: the three leading giants already secured 930 billion yuan, and companies like Zhipu and Minimax have gone public, leaving fewer opportunities for the remaining 200+ model startups. It is expected that many large-model companies will lay off employees, transform, or be acquired. Those that survive will either have niche applications (e.g., AI pharmaceuticals, AI programming) or backing from giants; pure “general-purpose large models” as a startup focus are no longer viable.
Overall, 2026 has been a year of “capital frenzy” in the AI sector, but beneath the surface, differentiation is accelerating: leading companies are receiving substantial funding, while smaller ones must either find niche areas or face elimination. Changes that consumers may notice include an increase in humanoid robots, more widespread use of AIGC-generated content, and potential faster development of new drugs through AI in pharmaceuticals—all results of heavy capital investment.