From Model Wars to People Wars
Forward deployed engineers are technical teams embedded directly inside enterprise customers’ operations to co-design, build, and run AI systems, focusing on security, integration, and day-to-day workflows so organizations can move from pilots to production-scale deployments with measurable business outcomes instead of experimental proof-of-concept projects. The big story is that the center of gravity in enterprise AI has shifted: success now depends less on who has the most powerful model and more on who can embed the best engineers. Amazon Web Services (AWS), Microsoft, Anthropic, and OpenAI have all decided that access to foundation or frontier models is no longer the real bottleneck. The bottleneck is turning those models into reliable, governed systems inside messy legacy stacks. Vendors that stay product-first will lose out to those that send people into the trenches.
AWS’s $1BN Bet on Embedded AI Engineering
AWS has moved aggressively, creating a dedicated Forward Deployed Engineering (FDE) organization designed to embed engineers directly inside customer teams. Backed by a USD 1BN (approx. RM4.6B) investment, this AWS engineering organization is built around an "agentic-first" development lifecycle, compressed timelines, and the goal of leaving customers self-sufficient when engagements end. These forward deployed engineers work alongside customer engineering, business, and security teams to build production AI systems using the customer’s own data, governance, and processes, compressing project timelines from months to days. Unlike traditional consulting, AWS says FDE is structured around shared goals and outcomes rather than billable hours. That is a direct challenge to classic vendor AI support models: instead of selling tools and distant advice, AWS is offering in-house capability building as part of its enterprise AI deployment strategy.
Microsoft, Anthropic, and OpenAI Join the Forward Deployment Race
AWS is not alone. Microsoft has introduced Microsoft Frontier Company, investing USD 2.5BN (approx. RM11.5B) and embedding 6,000 industry and engineering experts at customers to co-design, deploy, and continuously improve AI systems at scale based on business outcomes. Anthropic has formed a standalone enterprise AI services company backed by major investors, combining its Claude model engineers with consulting skills to integrate AI into core operations. OpenAI has launched the OpenAI Deployment Company, an enterprise FDE firm built by acquiring 150 specialists from applied AI consultancy Tomoro. These moves show that vendors are competing on forward deployment strategy, not only on model capabilities. The message is blunt: enterprises buying Claude, GPT, or any other model increasingly care whether vendors bring forward deployed engineers who will stay on site to solve implementation pain, not just ship an API.
Why Enterprises Need Hands-On Engineers, Not Just Better Models
Foundation models are widely available, and developers are chasing frontier models, but deploying AI in real enterprise environments is still hard work. Teams must integrate AI with existing systems, prepare proprietary data, set up governance, and meet security and compliance requirements before any application can go live. That implementation challenge is driving demand for forward deployed engineers who work directly with customer teams. According to Francessca Vasquez, Vice President of Frontier AI Engineering and Services at AWS, many customers "need expert AI engineers working directly with their teams to help them build and become AI-native organizations." Enterprises are no longer evaluating vendors solely on model quality but on whether they can implement AI securely, govern it effectively, and deliver measurable outcomes. In other words, deployment expertise now matters as much as model performance, especially where sensitive business and client data is involved.
From Proof of Concept to Production: What Comes Next
The wave of FDE investments signals a new norm: forward deployed engineering is becoming a standard piece of enterprise AI go-to-market strategies. AWS designs its engagements so customer engineers move from observers to co-builders to autonomous operators, leaving with deployed systems, knowledge graphs, runbooks, and architectural documentation anchored by a semantic layer in their own account. Microsoft talks about building an "intelligence platform" that compounds proprietary data and workflows over time, requiring enterprise AI engineering expertise with deep industry knowledge. Anthropic’s applied engineers plan to stay alongside customers over the long term. These vendor AI support models recognize a simple truth: enterprises need hands-on help to move from proofs of concept to production-scale AI systems. The winners in enterprise AI will be those who do the hard implementation work on-site, not those who only publish impressive benchmarks from afar.
