Summary of Key Points
Through more than a dozen days of on-site observation in Silicon Valley, the author reveals the true landscape of the AI revolution: B-side AI applications (especially enterprise services and coding tools) are experiencing explosive growth. Agent-based startups have shifted from selling tools to delivering tangible results, with even the emergence of infrastructure specifically for AI. While AI is reshaping business processes, it also brings concerns such as the ARR (Annual Recurring Revenue) bubble and job displacement. The 2B markets in China and the US are following different paths due to differences in labor costs and standardization of demand. Hardware companies, on the other hand, are reaping benefits from technological barriers. Overall, AI in Silicon Valley is currently focused on reducing costs and increasing efficiency for businesses; while disruptive innovations on the C-side (consumer market) have not yet emerged, the future holds promise.
1. How Profitable Are Silicon Valley AI Companies?
The growth of AI companies in Silicon Valley is not based on hype but on solid revenue. For example, Anthropic (the parent company of Claude) had annual recurring revenue (ARR) of only $9 billion at the end of 2025, which more than quadrupled to over $40 billion by May 2026, with coding tools contributing significantly to this growth. Investors initially underestimated the size of the coding market, thinking it was only worth $10 billion; yet Anthropic alone has far surpassed that estimate, shattering these assumptions.
The valuations of these companies are also quite realistic: Anthropic is valued at nearly one trillion dollars, which corresponds to 25 times its ARR. Considering its growth rate and future profitability, this valuation is much more reasonable than those of domestic AI companies that rely on narrative (such as Zhipu and Minimax). Even smaller companies focused on model aggregation can raise funds with valuations in the billions—because they have actual revenue, and their discussions revolve around how to increase revenue, not just creating grand visions for the future.
2. Agent-Based Entrepreneurship: From Selling Tools to Providing Direct Solutions
Silicon Valley is no longer talking about the domestically popular “local intelligent agent frameworks” but focusing on “Agents”—intelligent entities that can not only perform tasks but also deliver results directly. For instance, startups used to sell legal agent tools to law firms; now, they create Agents that handle legal tasks for companies outright (delivering tangible outcomes, which could lead to the elimination of entire departments).
More interestingly, there is a new trend in entrepreneurship: products are designed not for humans but for AI, featuring a “headless” design—no user interface or registration required, with only API interfaces for AI to use. Exa, for example, develops search engines for AI (whereas Google’s interfaces are designed for humans). It serves over 5,000 clients, including Cursor, Devin, and Alibaba. This shift in business models challenges the traditional model of “free content → attracting attention → selling ads,” as AI can now be paid based on traffic, potentially sharing revenue with content creators.
3. AI is Reengineering Business Processes
AI is not just about adding tools; it aims to transform entire organizational processes. For example, the two post-2000 founders of Corgi AI bought an insurance company and immediately eliminated all underwriters, using AI to reengineer the underwriting and approval processes. Traditional underwriters required decades of experience, but now AI can handle these tasks efficiently. This company is valued at $1.3 billion with annual recurring revenue exceeding $40 million, offering insurance policies to startups faster and more cost-effectively than traditional firms.
Investors describe this change as akin to “inserting an electric motor into a steam engine”: previously, product development involved months of design by project managers, coding by engineers, and testing by quality assurance teams; now, using tools like Claude Code, this process can be compressed to just two or three weeks. However, this efficiency improvement also creates bottlenecks (such as the need for continuous testing). While it may lead to job displacement for some employees, it also opens opportunities for new companies.
4. Differences Between Chinese and American 2B Markets
The prosperity of AI in Silicon Valley is driven by the B-side, while 2B entrepreneurship in China faces challenges due to high labor costs. In the US, companies prefer to buy tools rather than hire additional staff; in China, where labor is cheaper, large firms tend to build their own teams, leading to non-standardized requirements. As a result, startup products must be customized for each client, limiting their scalability.
US 2B startups focus on refining niche markets (e.g., legal agents) and quickly scaling by serving multiple innovative clients. In China, companies rely on custom projects and rely on project-based revenue without the benefit of compound growth. A Chinese entrepreneur once said, “Don’t try 2B in China—it’s too difficult.”
5. Concerns and Opportunities with the AI Revolution
Despite the progress, there are also challenges:
- ARR Bubble: Many companies are focusing on ARR, but many have negative profit margins or simply buy services from each other to boost their figures.
- Job Displacement: AI can increase efficiency by 3-5 times, reducing the need for certain roles (e.g., underwriters).
- Hardware Benefits: While the software sector is highly competitive, hardware companies (e.g., semiconductor equipment and materials) benefit from technological barriers. They enjoy the benefits of AI infrastructure, with major firms investing in computing power, allowing them to work remotely and see rising stock prices.
The author concludes by drawing a parallel with the pirate flag at the Computer Museum: AI may be like the Macintosh team in 1983—the product was not perfect, but it could change the world. Disruptive innovations on the C-side are yet to emerge, but everyone is working towards that moment.
Conclusion
Currently, AI in Silicon Valley dominates the B-side market, focusing on cost reduction and efficiency improvement. The “iPhone moment” for the C-side has not arrived, but entrepreneurs are riding the wave with both opportunities and anxiety. With persistence, perhaps the next disruptive team will emerge from an unsuspected place, shaping the future.
(The entire analysis is written in plain language to make complex financial and business concepts accessible to a general audience.)