Hard technology, which has been "out of favor" for many years, has become the "sweetheart" of capital again.
Throughout the first half of 2025, AI startups attracted more than 50% of the global venture capital funds.
Just a few years ago, the strategic investment departments of large domestic Internet companies were shrinking continuously, but they have become active again this year. As of now, Alibaba Cloud's strategic investment department has publicly announced three investment events, all basically centered around AI. Meanwhile, a host of Internet giants such as JD, Tencent, Meituan, Baidu, and ByteDance are also sparing no expense in investing in AI.
In addition, Legend Capital, Haier Capital, and some local funds have all shown great enthusiasm for AI.
According to IT Juzi data, Legend Capital has made 86 investments in the domestic artificial intelligence industry, involving 57 AI projects. Since 2024, there have been 7 local industrial funds that have made three or more investments in the AI field, namely Zhongguancun Science City, Zhong'an Capital, Wuxi Jintou, Sugao New Venture Capital, Shunxi Fund, Lingang Science and Technology Innovation, and Beijing Artificial Intelligence Industry Fund.
Globally, since 2022, the proportion of investment events in the AI field to the total number of investment events has been increasing year by year, rising from only 13.1% in 2022 to 32% in 2024. Overseas, Google alone has made 1,586 external investments since 2014, with 271 in the AI field, accounting for 17.1%.
An interesting phenomenon is that the entire investment circle's layout in AI has further extended to the entire AI industry chain. It permeates every aspect of the complete ecosystem, from upstream infrastructure, mid - stream platforms and production services to downstream applications. Driven by AI technological breakthroughs and industrial upgrades, hard technology has once again become the core battlefield for global capital competition.
The last time capital attached such importance to hard technology was 10 years ago when the concept of "smart hardware" began to gain popularity.
Capital "Competition" Squeezes Small and Medium - Sized Enterprises
In fact, the current AI track is already overcrowded. Data from the startup side shows that there are more than 1.9 million AI - related enterprises in China, and over 500,000 new ones have been added since 2024.
It is worth noting that when capital is maneuvering among numerous enterprises, it mostly focuses on large - scale enterprises.
According to IT Juzi data, in 2024, the number of investment events in the domestic AI industry was 500, a year - on - year decrease of 16%, and the total investment amount was approximately 66.316 billion yuan, a year - on - year decrease of 5%. Most of the funds flowed to leading large - model companies such as Zhipu AI, Baichuan Intelligence, and Yuezhi Anmian.
The Economic Observer estimated that in 2024, six leading large - model companies received more than 25 billion yuan in financing in total. In 2025, this situation continued: in the first quarter of 2025, there were 351 financing cases in the domestic artificial intelligence field, with a total investment amount of over 20 billion yuan. Among them, the top 20% of startups took nearly 80% of the financing.
Why does capital favor AI?
The most direct reason is to make money. Take "Qingzhi Capital" established by the Institute for AI Industry Research at Tsinghua University as an example. Data shows that from 2022 to 2024, Qingzhi Capital invested in nearly 20 AI projects, and its overall return on investment was more than three times the industry average. It was once rated as one of the "investment institutions with the fastest improvement in the industry in 2024".
However, when delving deeper into the underlying logic of this investment feast, it can be found that capital has different focuses.
First of all, it is not difficult to see from the investment stage that most capital is no longer as "bold" as before. Compared with betting on the seed round and angel round, capital prefers mature and initially - scaled enterprises, mainly focusing on mid - to - late - stage investments. A typical example is Tencent. Data shows that Tencent has more post - Series B and strategic investment events in the AI field. Series A investments account for 25%, and angel - round investments account for only 2%.
Globally, there is also a significant gap between the funding amounts in the seed round and Series A and those in the later Series C and D. Data shows that in the AI track, the median financing amount in the seed round is approximately $15 million (with an average of $41 million), and the median financing amount in Series A is approximately $75 - 80 million.
In contrast, the median financing amount in Series C and D is concentrated between $250 million and $300 million.
This is understandable. On the one hand, as the global economic environment continues to change, capital will become more and more cautious. On the other hand, the AI track has not yet emerged from the dilemma where the "burn - rate is higher than the earning - rate". Even Wall Street analysts once predicted that by 2026, large technology companies would spend $60 billion annually on developing AI models and only earn approximately $20 billion in revenue.
Secondly, although AI investment is spreading across the entire industry chain, the application layer always dominates. Data shows that in 2024, investors invested $4.6 billion in the application layer throughout the year, nearly eight times the $600 million in the previous year. Among them, more than half (53%) of Alphabet (Google)'s investment events were concentrated in the AI application layer.
Embodied intelligence and AIGC are the most common projects in the investment circle.
To date, application implementation has become a crucial part of the AI industry, which is also the main reason for capital's focus. In 2025, the application rate of global AI tools in fields such as code generation and customer - service robots has exceeded 25%. The research department of Huatai - Peregrine Fund pointed out that in terms of commercial implementation, AI applications are focusing on the B - end (enterprise - level) and C - end (consumer - level).
In addition, whether an AI enterprise can obtain financing is actually related to its "background". Local funds, in particular, attach great importance to this.
Among the local funds investing in the AI track, those from Beijing account for a large proportion. For example, Zhongguancun Science City, Shunxi Fund, and Beijing Artificial Intelligence Industry Fund participated in 4, 6, and 4 investment events respectively. At the same time, when examining major AI startups, their founders are basically related to Beijing universities. According to incomplete statistics from IT Times, there are already as many as 40 AI company founders with a Tsinghua background，and more than a dozen of them are relatively active.
Factors such as scale, profitability, capabilities, and direction have pushed the global AI capital feast to a climax. However, while the giants are frantically increasing their investment in computing power, data, and talent, from the release of ChatGPT to July 2024, 78,612 newly registered AI enterprises in China have disappeared.
With capital competing fiercely, the survival space of small and medium - sized enterprises is inevitably being compressed.
The "Industrial Capital Dream" in the AI Era
Among the capital investing in AI, industrial capital backed by giants seems to be the most active, and its enthusiasm has even surpassed that of vertical VC or PE institutions. According to IT Juzi data, from January 1 to August 27, 2024, there were 317 financings in the domestic AI field. Among them, there were 38 investment institutions that made three or more investments.
Among these 38 institutions, investment institutions with an industrial background accounted for more than 35%, totaling 14, overwhelming 13 VC institutions, 7 local industrial funds, and 4 PE institutions. Among industrial capital, Internet - based and leading AI - based entities have jointly staged a series of crazy investment behaviors.
On the Internet side, almost no company such as Alibaba, Baidu, Tencent, JD, and Meituan is absent. On the AI enterprise side, there are Zhipu AI, iFlytek and iFlytek Venture Capital, SenseTime and SenseTime Guoxiang Capital... Even not only leading AI companies, but in the past two years, AI enterprises that were once "invested in" are now mostly making investments.
Data shows that nearly a hundred domestic early - stage AI companies have invested in the AI industry since 2020, including dozens of listed AI companies such as Intellifusion, Cloudwalk Technology, and Horizon Robotics, as well as ten AI unicorn enterprises such as Biren Technology, Momenta, Saimo Technology, Pudu Technology, PlusAI, and Aibee Intelligence.
It should be noted that compared with the goal of investment and cashing out of traditional VC institutions, industrial capital often makes investments to improve its entire AI ecosystem, including technological synergy, resource sharing, application construction, and model innovation... Especially for Internet companies, when traffic growth hits the ceiling, it seems necessary to open up a second growth curve through AI.
Specifically, Tencent's investment department has been intensively deploying in the AI industry in recent years. It is not only for financial return considerations but also to inject AI into the corporate ecosystem. Multiple data shows that based on continuous investment in computing power, a powerful product matrix, and rich application scenarios, AI technology has begun to significantly drive Tencent's core business segments.
Similarly, Alibaba is trying to build an integrated intelligent ecosystem of "cloud - edge - end" by constructing AI computing - power infrastructure. ByteDance is aggressively testing commercialization scenarios in the field of generative AI, and Baidu Apollo is extending from autonomous driving technology R & D to intelligent transportation systems.
For leading AI enterprises:
Investment behavior first stems from the need to make up for technological shortcomings. In the context of the rapid iteration of AI technology, it is difficult for any enterprise to cover all technological dimensions on its own. Strategic investment seems to be the most efficient way to make up for the ability boundary. When computing power, algorithms, and data constitute the three pillars of the AI industry, investment can achieve the optimal allocation of these three elements the fastest.
Take Zhipu as an example. Previously, Zhipu spent hundreds of millions of yuan to acquire Lingxin Intelligence because this company has rich research experience in natural language processing (NLP), dialogue systems, natural language generation, and sentiment analysis. SenseTime participated in the financing of four companies, namely Tezign, Lingyun Universe, Artificial Productivity, and Youdi Technology.
Among them, Tezign started with technology.
Data shows that in 2024, Tezign published nearly a hundred academic papers, more than 20 of which were accepted by top - tier conference journals such as CVPR, NeurIPS, and IEEE T - PAMI. It has accumulated more than 1,440 valid patents and software copyrights. It is worth mentioning that Tezign's spatial intelligence system has been verified in multiple implemented projects.
Secondly, the intensifying competition in the ability to implement AI scenarios has increased the urgency of ecological investment. As of now, AI technology has entered the commercial verification stage. Leading enterprises can obtain scenario resources in vertical industries through investment to further improve their application ecosystem structure.
For example, Beijing Business Today found that Zhipu AI favors large - model application companies in the medical, legal, and cultural and entertainment industries; iFlytek Venture Capital focuses on three integration fields: AI + new hardware, AI + energy revolution, and AI + life science; SenseTime promotes the application of multi - modal large models in intelligent hardware such as robots, smart glasses, and intelligent vehicle cabins...
Since AI entered the integration stage after the "hundred - model war", the importance of ecological synergy has become more prominent. Perhaps for this reason, for the giants eager to embark on the AI path, investment is no longer just a simple capital transaction.
Is a Healthy Capital Ecosystem Urgently Needed?
The PitchBook report shows that in the first quarter of 2025, the total value of global venture capital transactions reached $126.3 billion (approximately 918 billion yuan), a year - on - year increase of approximately 53.46%, but the number of transactions was 7,551, a year - on - year decrease of approximately 32%. Among them, the AI market was the most active, and the transaction value accounted for 57.87% of the total global venture capital.
It is undeniable that AI investment shows characteristics of "irrational prosperity".
First of all, the AI investment boom is not a recent phenomenon. From 2016 to 2018, there was an overly optimistic investment wave in the computer vision and autonomous driving tracks. At that time, many startups received high valuations before forming a stable profit model, and the subsequent results were predictable.
This has led to many capital being "cheated". A typical example is Masayoshi Son and SoftBank.
In May 2025, the once - glorious Indian AI unicorn Builder.ai officially entered the bankruptcy liquidation process, and its $1.6 billion valuation disappeared overnight. This caused a huge setback for investors such as Masayoshi Son. In the past two years, SoftBank Vision Fund has been far from its previous glory, with a pre - tax loss of 115.02 billion yen (approximately 5.636 billion yuan) in the 2024 fiscal year.
In addition, data shows that from 2022 to 2024, more than 200,000 AI - related enterprises in China were deregistered or had their business licenses revoked, and a total of 353,000 have disappeared in the past 10 years. In August 2023, the net value of many domestic funds heavily invested in AI dropped significantly in two months. Among them, Huabao All - Connected fell by more than 26%.
More importantly, the global investment circle actually urgently needs a healthy capital ecosystem.
Since most investment institutions are concentrating on the hard - technology track dominated by AI, capital is overly concentrated, resulting in the investment in other strategic fields being continuously diluted. For example, as the electrification rate of terminal energy use continues to increase and the development of AI and data centers drives the growth of electricity demand, the fossil energy field has begun to be neglected by capital.
The World Energy Investment Report jointly released by the International Energy Agency (IEA), the Institute of Energy at Peking University, and the Energy Foundation shows that in 2025, upstream oil investment is expected to decrease by approximately 6% year - on - year, to approximately $420 billion, driving a approximately 4% decrease in oil and gas investment, which is the first decrease since 2020.
At the same time, investment in agriculture, water resources, and health is also decreasing. The World Investment Report 2024 recently released by the United Nations Conference on Trade and Development shows that the scale of investment in sustainable development goals has declined. The number of investment projects in fields such as agricultural food, water resources, and health is even less than that in 2015 when the sustainable development goals were proposed.
It is worth noting that in terms of the uneven investment, domestic local funds also have an obvious regional preference.
For example, local funds in Beijing have become the most active in the AI investment track, making Beijing's AI financing account for nearly 40%. In the first eight months of 2024, there were 107 AI projects in Beijing, accounting for 37%, far exceeding that of Guangdong (58 projects, accounting for 20%) and Shanghai (46 projects, 16%).
The preference of local funds for local enterprises may give rise to "migratory enterprises". In 2024, the number of core AI enterprises in Beijing exceeded 2,400, and the scale of the core industry was nearly 350 billion yuan, accounting for half of the country. In April 2025, as many as 26 enterprises were among the first to enter the new - quality AI industrial community in Beijing.
In fact, the rapid development of the global AI track is falling into a "technological iteration" dilemma, and the development speed also far exceeds the rhythm that the capital market can adapt to. When capital deviates from the fundamental and sustainable fields that require patient cultivation for a long time, the global economy will gradually lose the impetus for underlying innovation.
This article is from the WeChat official account "Consumer Frontline", author: