CITIC SEC: Leading model companies are forming FDE teams, and software companies may have opportunities for a significant value leap.

CITIC SEC published a research report stating that with the penetration of AI Coding, the establishment of a new AI commercialization main line is crucial. Model companies like OpenAI and Anthropic, which have achieved fruitful results in the coding field, are beginning to form FDE teams to seek deep integration between model capabilities and real business scenarios. The firm believes that software companies, as key players with customer positioning, delivery capabilities, and industry know-how accumulation, are expected to become important partners for model companies, seizing the value leap opportunities brought by FDE. Meanwhile, learning from Palantir, restructuring business models, establishing AI engineering delivery capabilities, and exploring outcome-based billing are important indicators of successful FDE transformation for software companies. It is recommended to pay attention to companies with potential and foundations for FDE business development, including: 1) model vendors; 2) enterprise software vendors; 3) IT service vendors; 4) vertical industry vendors. CITIC SEC's main points are as follows: FDE was pioneered by Palantir, delivering results deep into the business frontline, which in turn feeds back experience to promote platform evolution. FDE stands for Forward-Deployed Engineer, which is a compound technical position that brings engineering capabilities directly to the customers business site. The mission of an FDE is to bridge the last mile gap between standardized product capabilities and the complex business scenarios of customers, taking end-to-end responsibility for the technical implementation results on-sitecovering everything from requirement discovery and solution validation to production launch, user adoption, and achieving business results. In comparison, pre-sales solution engineers facilitate contract signing and are responsible for solution demonstrations and opportunity conversions; traditional outsourcing and on-site developers complete the contracted workload and are responsible for the delivered code and functionalities; FDEs address the core business issues of customers and are accountable for the final business outcomes. Palantir achieves high profitability and strong customer retention through FDE. Palantir employs an Echo-Delta collaboration mechanism internally: the Echo team consists of embedded analysts who engage on-site with customers to deeply explore business pain points and convert them into precise requirements; the Delta team's software engineers quickly build prototypes on-site and iteratively implement them for customers. Palantir initially engaged in G-end business, assisting government intelligence agencies in constructing counter-terrorism and intelligence analysis data systems; in 2016, it entered the B-end, servicing industries such as finance, healthcare, energy, and manufacturing; by 2023, FDE has entered a new phase: the delivery focus has shifted from traditional big data integration and dashboard building to deploying AI Agents and automated decision flows. At the same time, a substantial amount of basic configuration and code generation work can be supported by AI, greatly enhancing efficiency. Unlike outsourcing, FDE's on-site delivery is not solely based on human output; instead, FDEs write code on-site, and the back end extracts reusable components, solidifying them as platform functions. Based on a continually thickening platform foundation, the marginal delivery costs decrease with each new customer. Palantir maintains software-level high profitability, with the adjusted gross margin reaching 86% in Q2 2026, GAAP operating profit margin at 47%, and adjusted operating profit margin at 62%, setting a historical high. Leading model companies are forming FDE teams to deepen their partner networks. Leading model companies such as OpenAI and Anthropic actively seek new growth points beyond coding, with FDE expected to become a lever to deeply connect model capabilities with enterprise needs. According to a report from The Wall Street Journal, in May 2026, OpenAI established an independent subsidiary, OpenAI Deployment Company, which secured $4 billion in funding and acquired AI consulting and engineering firm Tomoro to gain around 150 experienced FDE specialists. They are currently screening the first batch of clients, prioritizing those with long-term cooperation relationships with OpenAI. OpenAI is also collaborating with consulting firms like BCG, Bain, and Accenture to enhance communication capabilities at the enterprise level. According to reports from Titanium Media, at the same time, Anthropic, in partnership with Blackstone, Goldman Sachs, and others, established a joint venture focused on enterprise AI services, securing $1.5 billion in investment commitments. Anthropic's Applied AI engineers will work with client teams to seek the best application scenarios for Claude and develop customized systems; Anthropic is also rapidly expanding its FDE coverage through partnerships with global consulting and IT service firms such as Accenture and UST. The firm believes that during the process of forming FDE, the importance of software companies and consulting firms will become evident due to resource allocation considerations. Software companies are likely to become key partners in implementing model capabilities, seizing opportunities for value elevation. As the problem of companies not knowing how to use AI or using it ineffectively becomes increasingly prominent, FDE demand is expected to surge. Gartner predicts that by the end of 2026, more than 85% of technology service providers will make FDE projects a core means of delivering AI and shortening deployment cycles. The firm believes that software companies possess significant potential to develop FDE: first, customer positioning. Software project-based companies have long-term service relationships with top government and enterprise clients in manufacturing and finance, holding the entrance to AI scenarios; second, delivery system. Software companies have well-established on-site management systems, making them easier to adapt to the FDE model; third, accumulation of industry know-how. Software companies are familiar with client business processes, system architectures, and data situations, enabling them to identify core problems and build systems. FDE also raises higher requirements for software companies: establishing a closed loop that ranges from on-site customization to commonality screening to product abstraction and ultimately achieves scalable reuse, building AI engineering capabilities such as fine-tuning large models and orchestrating Agent workflows, and exploring a shift from time-and-material billing to performance-based billing. The firm believes that software companies with deep industry customer positioning, partnerships with leading large models, and rapid establishment of FDE business models are expected to achieve a leap in commercial value. Risk factors include: model technology innovation and capability breakthroughs falling short of expectations; FDE business model construction and application scenario expansion not meeting expectations; AI commercialization deployment not reaching expectations; and intensified industry competition.