Who will become the Apple of the AI era?

When OpenAI released GPT - 4, Silicon Valley investor Andreas Hoffmann asserted, "We are standing at the starting point of a computing revolution." Three years have passed, and this statement has transformed from a prediction into reality.

However, when we extend our view to the historical dimension, this transformation seems more like a "long - overdue awakening." In 2012, AlexNet emerged in the ImageNet competition, quietly sowing the seeds of deep learning; in 2016, AlphaGo defeated Lee Sedol, hinting at the "extraordinary capabilities" of AI; in 2022, ChatGPT sparked public imagination, and capital, talent, and computing power began to pour in crazily.

Especially since this year, the venture capital field has witnessed a historic turning point: AI startups have captured 51% of the global total venture capital for the first time, exceeding the sum of all other fields. This data comes from the latest report of CB Insights, marking an unprecedented pursuit of artificial intelligence by capital. Among them, the United States holds an absolute leading position in the wave, contributing 85% of the AI financing and 53% of the trading volume.

Data shows that the global investment scale in the AI market is approaching $200 billion, but no company with "Apple - level" disruptive capabilities has emerged yet. Looking back at Apple's rise, when Steve Jobs founded Apple in 1976, personal computers were a "game for engineers," but the Macintosh redefined human - computer interaction with its graphical interface and user experience; in 1998, the iMac broke the "industrial design shackles" of color electronic products; in 2007, the iPhone transformed the mobile phone from a communication tool into an entrance to the mobile Internet. Every time it achieved a disruption, Apple completed a closed - loop from technological breakthrough to ecosystem construction.

Today, players in the AI industry are trying to replicate this path: Large models are the "Macintosh," hardware is the "chip," and applications are the "App Store". But the question is - in the "Cambrian explosion" of AI, who can find a balance among technology, ecosystem, and business and ultimately become the "Apple" of the new era?

01

AI Entrepreneurship: One Part Fire, One Part Water

Looking back at the AI entrepreneurship journey in the past three years, it can be clearly divided into three distinct stages.

From 2022 to 2023 was the foundation - laying period for large models. ChatGPT emerged suddenly, igniting the boom in generative AI, and global tech giants and startups all bet on the development of underlying models.

In 2024, it entered the application exploration period. As the technology gradually matured, application - layer tools such as Cursor, Midjourney, and Perplexity rose rapidly, marking the transition of AI from technology demonstration to practical value creation.

In 2025, it stepped into the vertical integration period. AI startups began to deeply embed themselves in various industries, seeking commercialization paths in specific scenarios.

Among the 169 startups at YC's Summer Demo Day in 2025, more than half of the projects took AI agents as their core direction. These companies no longer pursue general large platforms but instead strive to penetrate deeply into vertical fields, targeting those "jobs that people are reluctant to do, not good at doing, and are particularly expensive."

For example, Solva uses AI to automate insurance claims and achieved an annualized revenue of $245,000 within 10 weeks of its launch; Autumn specializes in solving the complex billing problems of AI companies and has been used by hundreds of AI applications and 40 YC startups. In the medical field, Perspectives Health generates medical records and forms in real - time by monitoring doctor - patient conversations, saving doctors half of their paperwork time and maintaining a 25% weekly growth during the pilot phase.

However, there are hidden concerns behind the prosperity. AI entrepreneurship is showing an obvious polarization: On one side, application - layer companies are booming, while on the other side, there are high barriers and resource concentration in the infrastructure field.

Indeed, data shows that the number of newly - added global AI unicorns actually decreased by 12.50% year - on - year and 6.67% quarter - on - quarter, indicating that the market is undergoing a structural adjustment. The domestic market is the same. From the early "Six AI Tigers" to the "Six AI Dragons" in Hangzhou, most enterprises have mediocre ecosystem and sustainable operation capabilities, and only a few can achieve large - scale revenue.

The attitude of the capital market has also become more rational. Investors no longer only focus on the novelty of technology but pay more attention to user retention, unit economic benefits, and computing power costs. That is to say, AI entrepreneurship is moving from the fanatical period of chasing hotspots to the structural adjustment period of value verification.

02

What Does AI Entrepreneurship Lack to Be Like Apple?

What makes Apple what it is today is not just the iPhone or MacOS but a set of "counter - intuitive" underlying logic.

First is strategic determination: From the iMac in 1998 to the iPhone in 2007, Apple took 9 years to upgrade "consumer electronics" to a "lifestyle brand"; second is the closed - loop ecosystem: The "hardware - software - service" iron triangle composed of the App Store, AirPods, and Apple Watch makes it difficult for competitors to imitate; finally is organizational resilience: The "paranoid" culture of Steve Jobs and the "operational philosophy" of Tim Cook complement each other, ensuring the company's balance between innovation and profitability.

In contrast, the current AI industry has three major shortcomings that restrict the emergence of "Apple - level" enterprises: First, the separation of technology and business: Large - model developers and hardware manufacturers lack ecological synergy, resulting in the inefficient transformation of technology into products; second, the lack of organizational capabilities: Most AI companies still stay in the "engineer thinking" and ignore user experience and brand building; third, the mismatch of the capital cycle: Venture capital chases short - term hotspots excessively and ignores long - term infrastructure.

That is to say, most AI startups still stay in the stage of "tool providers" and have not formed a real closed - loop ecosystem.

Specifically, domestic AI entrepreneurship shows a generational change from the "Six Tigers" to the "Six Dragons." The early "Six AI Tigers" got stuck in the quagmire of losses in recent years due to their excessive reliance on the to - B scenario; while the new generation of entrepreneurs target the to - C track, such as AI writing and code - generation platforms. However, these enterprises also face challenges - how to survive under the ecological blockade of giants and the impact of open - source models?

According to Gartner statistics, among global AI startups in 2023, 62% of the products were iterated more than 3 times within 18 months, but only 17% could achieve a positive commercial cycle. This reveals a cruel reality: The essence of AI entrepreneurship is a game of "computing power leverage" - whoever can find the optimal solution among model performance, data quality, and cost control will survive.

From the changes in the investment market, we can also see the stage characteristics of the AI industry. Data shows that in the third quarter of 2025, the global total venture capital reached $95.6 billion, but the number of transactions dropped to the lowest level since 2016. This indicates that investors are becoming more picky and are investing larger - scale funds in more mature and high - potential projects.

03

A New Cycle and New Opportunities

Looking back at history, AI has experienced three waves: In the 1980s, expert systems: They were short - lived due to the lack of data and computing power; in the 2000s, machine learning: It relied on manual feature engineering and failed to break through the "black - box" dilemma; in the 2020s, large models: They achieved general intelligence through self - supervised learning and massive data, but their implementation is still restricted by scenarios.

Compared with the previous two AI waves, the transformation brought about by this large - model revolution is indeed more drastic, with a "double - helix structure," manifested in the simultaneous explosion of technological breakthroughs (large models) and industrial demands (digitization). According to IDC data, in the first half of 2025, the scale of the Chinese AI IaaS market soared by 122.4% year - on - year, reaching 19.87 billion yuan. The growth rate of the GenAI IaaS market was even as high as 219.3%.

In addition, the development of AI at home and abroad also shows different characteristics.

The overseas market is driven by the innovation of basic models. Companies such as OpenAI and Anthropic are constantly raising the upper limit of model capabilities; the Chinese market focuses more on application implementation, relying on a large user base and rich scenario resources to promote the commercialization of AI.

If analyzed from an industry perspective, the fundamental problem lies not in the imagination of the business itself but in the change of the supply - demand relationship. On the one hand, the supply of computing power is becoming more diversified. Cloud providers at home and abroad are deploying self - developed chips, and the resource supply and price in the computing power market are generally stable. On the other hand, the demand structure is being reshaped. Enterprises are no longer satisfied with simple model training but are more concerned about how to integrate AI capabilities into business processes to achieve a value - closed loop.

Against this background, some emerging forces are quietly rising.

Taking chip design as an example, the revenue of Hygon Information increased by 54.65% in the first three quarters, and the revenue of Cambricon soared by 2386.38%, demonstrating the potential of domestic AI chips. LiblibAI, which focuses on AI visual creation, completed a $130 million Series B financing, becoming the largest single - round financing in the domestic AI application track, indicating the capital market's re - evaluation of application - layer companies.

In other words, AI is shifting from being driven by "resource supply" to being driven by "innovation empowerment."

To some extent, the emergence of Apple was the result of the "madness" of Steve Jobs in the garage in 1976 and the "rationality" of Tim Cook when he took over in 1997. As for who will become the "Apple of the AI era"? Perhaps the answer lies in these keywords: long - termism, ecological thinking, and user - centricity. Just as the iMac in 1998 broke the "dull tradition" of electronic products with rainbow colors, future AI companies must find their own "breaking point" - not competing on who has larger model parameters but redefining the relationship between humans and the world with technology.

This article is from the WeChat official account "New Vision" (ID: xinmouls), author: Tang Ning, published by 36Kr with authorization.