China Galaxy Securities has released a research report, stating that FDE is expected to drive cost reduction, efficiency gains, and enhanced commercial value for AI products in enterprise settings. This will accelerate the deployment of AI applications and AI agents within businesses, prompting a value reassessment for FDE service providers and AI application developers. The report suggests focusing on: 1) FDE service providers; 2) AI + office software; 3) AI + marketing; 4) AI + finance; 5) AI + industrial software; 6) AIGC. Industry-specific expertise will become a key moat for future AI application companies, and some software firms may see a reassessment of their value systems, unlocking new valuation ceilings.
FDE Solves the "Last Mile" of AI Application Deployment, Acting as a Deep AI Delivery System
FDE, or Forward Deployed Engineer, involves deploying engineers directly to the front lines of a business. Through on-site exploration and rapid iteration, this model integrates large language models and other AI technologies into a company's internal business processes, achieving cost reduction, efficiency gains, and real business value. This concept was first proposed by the US data analytics company Palantir and is now widely adopted by major firms like Microsoft, OpenAI, and Anthropic. FDE is not a new profession but an upgrade of the solution architect role in the AI era. Its core value lies in penetrating a company's surface-level needs and breaking them down into actionable solutions. Required skills include full-stack software engineering capabilities, along with product and commercialization thinking. In essence, FDE is a system for deeply delivering AI applications and agents. It can be categorized into product-driven FDE, vertical-scenario FDE, and pure project-customized FDE.
Accelerating AI Industry Transformation Propels FDE as a New Paradigm for AI Application Delivery
Recently, Beijing released the "Several Measures on Accelerating the Development of Intelligent Agents," which explicitly supports model service providers and industry enterprises in using innovative models like FDE to accelerate the deployment of AI agent applications. The report believes the rise of the FDE model is not accidental but a product of the AI industry's transformation. The continuous iteration of domestic large models like Kimi and DeepSeek, along with the emergence of AI agents, has lowered the barriers to enterprise application, accelerating intelligent transformation across industries. FDE can address three key pain points companies face with AI: 1) Highly non-standardized business processes: Many legacy systems like ERP, OA, and MES create significant data silos with hidden business rules that general-purpose large models cannot adapt to. FDE is needed on-site to break down these processes and write customized code. 2) Data security and privacy barriers: Highly sensitive sectors like government, finance, and military have strict data sensitivity. Public cloud APIs pose risks of core data leakage. FDE must work within internal network environments to complete the entire closed loop, from model private deployment and data governance to business model design and customized agent delivery. 3) Implementation effectiveness and ROI evaluation: The comprehensive capabilities of current domestic base models have crossed the threshold of usability, with fundamental performance sufficient to support industry deployment needs. Companies are now more focused on how to fully leverage the capabilities of large models and the ROI they bring. Furthermore, overseas tech giants have already taken the lead in deploying FDE. Microsoft has established a large, independent FDE department, Anthropic has partnered with Blackstone and Goldman Sachs to form a joint enterprise service venture, and OpenAI has upgraded FDE to an independent, major business line. These moves have fully validated FDE as an essential path for the practical deployment of AI applications.
FDE Poised to Become a Standardized Path for AI Deployment, Software Sector May See a Value Reassessment
In the era of large AI models, FDE is no longer a shallow form of traditional human outsourcing. Instead, it relies on forward-coding capabilities and business know-how to deeply embed the power of general-purpose large models, accumulating them into reusable enterprise data assets and intelligent platforms. Therefore, traditional software industry revenue growth will no longer depend on conventional software licensing models. AI customization services, private deployment, model operations, and on-site engineering services will increase average customer spending. The report argues that FDE is bringing about a revolution in the delivery paradigm for the software industry, enhancing customer stickiness and average transaction value. The business model is shifting from traditional software licensing to a lifecycle of incremental value services focused on ROI and usage outcomes. Consequently, industry-specific know-how will become a key competitive moat for future AI application companies, and some software firms will face a reassessment of their value systems, potentially unlocking new valuation ceilings.
Risk Warning
Risks include: technology iteration falling short of expectations; intensified competition among tech giants; legal and regulatory risks; supply chain risks; and downstream demand falling short of expectations.