1. [para. 1][para. 2] The success of AI companies and their surging valuations depend on delivering practical enterprise applications, according to Bain Capital's Greater China Chairman Jonathan Jia Zhu. As valuations rise, driven by the unprecedented growth of foundational model developers, global capital focus is shifting from infrastructure to the application layer where long-term winners are expected to emerge.
2. [para. 3][para. 4] Valuations across the AI supply chain have surged in 2026, with OpenAI and Anthropic reaching $1 trillion each, and Nvidia and SpaceX hitting $4 trillion and $2 trillion respectively. Investors are shifting away from traditional valuation methods; growth-stage firms are assessed on cash flow and EBITDA, while mature firms are evaluated on net profit and free cash flow.
3. [para. 5][para. 6][para. 7] As the "Magnificent Seven" tech stocks slowed significantly this year, trailing the broader Nasdaq index, investors have sought opportunities in unlisted AI firms. These high valuations are driven by rapid growth and competitive edges, though it remains uncertain if these edges form stable long-term expectations. Investors now value high-growth model developers using annualized recurring revenue (ARR). OpenAI's ARR grew from $2 billion (2023) to $6 billion (2024) to over $20 billion (2025), while Anthropic's grew more than tenfold annually over three years. Zhu called maintaining such rapid ARR growth extremely rare but noted it provides a solid valuation basis, unlike the metrics used during the internet bubble era when companies were valued on user traffic without actual revenue.
4. [para. 8][para. 9] Private equity and venture capital firms show differing preferences for AI investments. As OpenAI and Anthropic shift toward enterprise-facing businesses, new opportunities have emerged for PE firms. Bain Capital and TPG invested in OpenAI Deployment Company, an OpenAI subsidiary focused on expanding its enterprise client base. Zhu said enterprise-level AI requires not only capital and computing power but also professionals familiar with different industries, making it a natural fit for PE firms that possess industry experience and talent pools.
5. [para. 10][para. 11][para. 12] Bain Capital has invested in AI infrastructure, including memory chipmaker Kioxia Holdings Corp., whose stock price jumped 350% this year driven by AI demand, data center operator Chindata Group Holdings Ltd. (where Bain eventually separated its Chinese business from overseas operations), and global optical module manufacturer Coherent Corp. The benefits of AI development have spread globally, driving stock market gains in South Korea and Taiwan and fueling Japan's recent rally through semiconductor supply chain companies.
6. [para. 13][para. 14] Zhu expects a large number of AI application companies to emerge in the future. Drawing a parallel to the history of internet-related businesses, the most valuable companies ultimately stood out in the application layer rather than infrastructure. The future of foundational model developers hinges on whether they can put their technologies into commercial application for business.
7. [para. 15][para. 16] Leading model developers like Anthropic are already generating significant enterprise-service revenue, demonstrating both platform and application attributes, while many new application tracks await exploration by startups. The ability of enterprises to put AI into practical use has become a key consideration for institutional investors. All companies in Bain Capital's portfolio are exploring using AI to empower their businesses and improve operational efficiency.
8. [para. 17] The development of AI also requires physical hardware, an area where China holds a clear development advantage, particularly in robotics and medical AI applications, according to Zhu.
AI generated, for reference only
Sign in with Google
Sign in with Facebook
Sign in with 财新