Financing raised within half a year is 1.7 times higher than the total amount of last year. Is there a bubble in the AI sector? Investors noted that more than 90% of the capital will be invested in the AI field in five years.
"In the first half of 2026, the total financing amount of the global AI market has exceeded 1.7 times that of the whole year of 2025." From September 9 to 12, the 2026 Inclusion·The Bund Summit was held in Shanghai. A set of data disclosed by He Fei, investor of Alibaba Entrepreneurs Fund AEF NextGen Fund, at the "Venture Capital Meetup" hosted by the summit confirms that the investment enthusiasm in the global AI sector continues to rise.
The "Venture Capital Meetup" of this year's Bund Summit attracted more than 500 investors from around the world and over 120 AI startups to conduct high-density exchanges on the Bund. Compared with the first "Venture Capital Meetup" set up at the 2025 Bund Summit, the project scale, capital density and track coverage of this year's event have been further upgraded.
A notable phenomenon is that the focus of on-site investors has changed significantly. In addition to paying attention to "whether the technology is leading", investors are more asking "whether the customers are real, whether the delivery can be replicated, whether the revenue can be sustained, and whether the team can survive through cycles".
At the same time, many entrepreneurs are also "reverse selecting" investors, hoping to find not only capital, but also investment partners who truly understand the industry. A founder of smart pet care said that the industry judgment and resources brought by investors may be more critical than a sum of money.
On September 11, a number of senior investors said in interviews with media including Time Weekly that AI is accelerating its evolution from a technical variable to an industrial variable, and capital's layout in the new AI economy is shifting from concept chasing to systematic investigation of real implementation capabilities.
"AI will penetrate into more and more new fields, and many traditional fields will be reshaped. From the perspective of the investment strategy of the entire fund, I believe that more than 90% of capital will be invested in the AI sector in 5 years," said Yan Qianhang, Vice President of Frees Fund.
AI Bubbles Are Not Completely Worthless, The Challenge Lies In Judging Whether Enterprises Can Survive Through Cycles
Under the upsurge of AI investment, bubble has become an unavoidable topic.
Zhai Ning, Founder of Mingshen Capital, believes that in terms of revenue and net profit, there are indeed bubbles in the current AI industry, and many companies do not yet have a stable profit model; but bubbles are not completely worthless. The key is to see whether they attract the most talented people to conduct real research and solve real problems, rather than just creating valuation illusions. In terms of founder portraits, Zhai Ning values "awe" more. Being smart is becoming an entry ticket, and the ability to understand regulation, law, ethics, customers and industrial interests determines whether a company can grow from 0 to 1 and then to 100.
Feng Tian, head of AI investment at BAI Capital, expressed a relatively optimistic view on this. She believes that bubbles are positively correlated with innovation. Without bubbles, a large amount of capital and talents will not pour in. The challenge for investors is to judge which companies can survive through cycles after the bubble bursts.
At the valuation level, Feng Tian said she still believes that the market is effective. In the process of innovation, it is unavoidable for investors to pay a certain premium, because both investors and entrepreneurs are on the same starting line, and investors also need to make quick decisions when facing completely unprecedented things.
Jiang Haonan from GGV Capital reminded that the valuations of primary market projects have risen recently, and the valuation of some teams in the first round has reached 3 to 5 times that of two or three years ago. He believes that in the short term, the secondary market valuation will face great challenges. On the one hand, overseas Anthropic and OpenAI will go public one after another, and SpaceX has already gone public, so a large amount of capital will be absorbed into these companies; on the other hand, several leading large model companies in China are also preparing for listing successively, and capital will also gather towards the leaders.
Jiang Haonan suggested that entrepreneurs stay rational, make each product reasonably correspond to a valuation, and raise corresponding funds to support business development. Primary market investment should be considered at least based on a 5-year cycle, and investors also need to judge whether enterprises can survive through cycles.
Embodied Intelligence Is Still In The Early Stage, And An Inflection Point May Appear In The Next 3 To 5 Years
The embodied intelligence track continues to heat up and has become one of the most popular directions in the AI venture capital field.
According to Wang Chao, AI investor and Executive Director of CAS Star, CAS Star has been laying out embodied intelligence since the first half of 2023, and first invested in Agibot, and then successively invested in some teams with autonomous driving background.
He said that CAS Star believes that embodied intelligence is a long-term track, not a short-term speculative behavior, and will continue to pay attention to and lay out this field in the future, focusing on two types of companies: one is companies with strong system capabilities and focusing on implementation, and the other is teams and companies that can continuously break through the ceiling of embodied intelligence.
However, in the eyes of investors, embodied intelligence as a whole is still in a very early stage at present.
Yan Qianhang believes that embodied intelligence is still in the early R&D stage, and technological development will not grow linearly every year as expected, and qualitative change will not occur until a relatively favorable inflection point is reached. He said that the current investment environment in the primary market is relatively comfortable, because under the background that the technology has not reached the inflection point, investors can continuously explore new ideas and find teams with the highest R&D efficiency.
As for the commercial implementation issue, Yan Qianhang believes that it is too early to discuss it now, which is similar to the situation in 2012 when people could not see the clear implementation scenarios for autonomous driving, and we need to wait until the technology has the mature capabilities to meet commercial benchmarks before further observation.
Wang Chao also believes that embodied intelligence is not ready yet in either the training paradigm or the data dimension. "From the perspective of the whole industry, the amount of money each company gets is instead insufficient, or even far from enough. So I think this field definitely needs continuous investment and layout. There are bubbles in some parts, but the overall amount of capital is insufficient," said Wang Chao.
Zhao Fuheng, Investment Director of Jiuhe Venture Capital, divides the bubbles of embodied intelligence into two categories: one is that the expected commercial value of solving problems in the future is greater than the actual commercial value that can be solved, and the other is that a large number of successful enterprises are expected to emerge in the market, but in the end only a few of them can stand out. Zhao Fuheng believes that the former is difficult to prove or disprove, while the latter does have bubbles to a certain extent from the perspective of quantity.
"If we lower our expectations, the industry bubble actually does not exist. Everyone is still moving forward step by step, and there is a chance to see an inflection point in the next 3 to 5 years," said Yan Qianhang.
This article is from the WeChat official account "Time Weekly" (ID: timeweekly), written by Huang Yukun, edited by Xi Ang, and published with authorization by 36Kr.