CITIC SEC has released a research report stating that leading model companies are accelerating the penetration of AI into deep industry scenarios, promoting enterprise-level AI adoption. Internationally, Anthropic and OpenAI have established FDE joint ventures to deeply embed AI into core business processes. Domestically, according to a disclosure by MiniMax on May 11th, the company officially launched its '10x Team' initiative, mirroring collaborative models like Anthropic Fellows and Google DeepMind. This aims to strengthen cooperation with top experts across various fields, striving to replicate the 10x efficiency leap seen in programming across more industries. AI commercialization may be approaching an inflection point, with enterprise application deployment expected to accelerate. It is recommended to monitor the Annual Recurring Revenue (ARR) growth trends of leading model companies and subsequent progress in model applications.
The key points from CITIC SEC's report are as follows: On May 4, 2026, leading AI companies Anthropic and OpenAI announced on the same day the establishment of their respective joint ventures with top-tier private equity funds. These ventures aim to provide enterprise clients with on-site engineer services (FDE) similar to Palantir, driving the deep integration of AI into core business operations. Anthropic's partners include anchor investors Blackstone, H&F, and Goldman Sachs, along with five co-investors such as Sequoia Capital. OpenAI, in collaboration with 19 investors including TPG, Bain Capital, and SoftBank, jointly established The Deployment Company (DeployCo). This simultaneous strategic move by the two AI leaders signals that competition in large language models is extending beyond mere model performance to a comprehensive contest of enterprise-level productivity transformation capabilities.
OpenAI secures control with guaranteed returns, while Anthropic opts for shared risk. According to Bloomberg, OpenAI's led joint venture (DeployCo) has an initial scale of approximately $40 billion with a valuation reaching $100 billion. OpenAI is contributing $1.5 billion ($500 million initially, with an additional $1 billion to follow) and has promised its private equity investors a locked-in five-year period with a guaranteed annualized return of 17.5%. This arrangement allows OpenAI to maintain strategic control over the joint venture through super-voting rights. In contrast, the joint venture established by Anthropic is relatively smaller, with an initial scale of about $15 billion, and employs a common equity model, sharing risks with its investors.
The FDE model directly addresses the pain point of difficult enterprise-level AI deployment. Current enterprise demand for AI is robust: according to Epoch AI, Anthropic's ARR has surpassed $44 billion, with over 1,000 enterprise clients spending more than $1 million annually. As of January 2026, enterprise client revenue accounted for 40% of OpenAI's total, with expectations to rise to 50% by year-end. However, a business model relying solely on API sales struggles to drive AI deep into core business processes. Enterprise clients urgently need customized integrations built around their proprietary data, legacy systems, and internal workflows. The FDE model, which stations engineers on-site at client companies for close collaboration, enables timely responses to personalized needs and a deep understanding of complex business logic, thereby facilitating highly tailored AI deployments for core business processes. Once an AI system is deeply embedded in core operations, the high replacement cost and strong client stickiness can generate long-term revenue for the AI company. To this end, OpenAI announced the acquisition of AI consultancy Tomoro on May 11th, bringing approximately 150 FDE engineers and deployment specialists into DeployCo.
Private equity partners provide vast customer channels and reduce internal decision-making barriers. The joint ventures established by OpenAI and Anthropic target mid-sized enterprise clients who have AI application needs but lack in-house AI engineering teams or budgets for self-build projects. Private equity funds not only offer financial support but also bring extensive customer access: 1) PE firms have numerous portfolio companies, and partnering with them helps AI companies gain priority access to enterprise clients within their ecosystem. 2) As shareholders, PE firms exert considerable influence over their portfolio companies, enabling them to facilitate cooperation with AI companies from the top down, reducing internal decision-making obstacles and significantly shortening approval cycles. 3) PE firms benefit from sharing in the joint venture's profits while also enhancing the value of their portfolio companies through AI empowerment, creating a natural incentive to foster such partnerships.
Developing FDE business through joint ventures allows for a rational valuation framework. 1) The gross margin for FDE services is approximately 30-50%, significantly different from the typically over 80% gross margin for SaaS APIs. 2) Capital markets typically assign valuation multiples of only 8-15x to 'software + services' hybrid companies, compared to 20-30x for pure software companies. Given Anthropic's current trend of rapid ARR growth, incorporating the FDE business directly into the parent company could potentially lead to a valuation gap in the hundreds of billions of dollars. As both Anthropic and OpenAI are currently in pre-IPO stages, establishing separate joint ventures to house the FDE business effectively insulates the parent company's valuation framework from its negative impact.
Risk factors include slower-than-expected AI application deployment, slower-than-expected iterations of foundational models, and a potential decline in enterprise users' willingness to pay.
Comments