What is a Forward Deployed Engineer (FDE)? Why AWS, OpenAI, and Anthropic are sparking an AI talent war
A Forward Deployed Engineer (FDE) is an engineer who works directly with enterprise customers, responsible for integrating AI models, data, and existing workflows into production systems. As businesses begin demanding revenue or cost benefits from their AI investments, AWS, OpenAI, and Anthropic have successively expanded their enterprise deployment services, and FDEs have quickly become talent that the AI industry is fiercely competing for.
Executive search firm Christian & Timbers estimates that demand for FDEs among US companies will increase by 2,100% by the end of this year. At the beginning of this year, only about 5% to 10% of surveyed companies were prepared to hire FDEs; by the end of Q2, that figure had risen to 70%.
Christian & Timbers estimates that there are currently about 17,000 FDEs in the US, but the total number of talent who simultaneously possess industry knowledge, enterprise communication skills, and applied AI experience—and can continuously deliver returns on investment for clients—may be only around 2,000.
The 2,000 figure refers to the total number of qualified talent across the US, many of whom are already employed—it does not mean there are 2,000 available candidates on the market waiting to be recruited.
What is an FDE? How does it differ from a regular software engineer?
The FDE working model has long been closely associated with data analytics company Palantir. Palantir calls FDEs Forward Deployed Software Engineers (FDSE)—engineers who go directly into customer organizations and use the company's software platform to solve specific business problems.
Palantir's distinction between the two roles is straightforward: a regular software engineer typically develops a product capability that serves multiple customers, while an FDE uses multiple product capabilities to tackle the complex problems of a single customer.
Take the FDEs that AI company OpenAI is currently recruiting as an example: the job typically covers exploring requirements, defining technical scope, designing systems, writing code, and deploying systems to production environments. Engineers must also work closely with customers' technical teams and industry experts to ensure the systems can connect to existing processes and keep running.
As a result, FDEs need software engineering, product development, project management, and industry knowledge all at once. Beyond understanding models and APIs, they also have to handle data formats, access control, information security, regulatory constraints, and internal operational processes.
The final deliverables of an FDE cannot stop at the demo stage either. What enterprises expect is a system that works in production, gets adopted, and whose results can be measured.
Why are companies suddenly scrambling to hire FDEs?
Over the past two years, the barrier for enterprises to access advanced AI models has gradually lowered. Multiple model companies offer enterprise plans and APIs, and the capabilities of open-weight models keep improving. The challenge facing enterprises has consequently shifted: the models have been bought—now how do you integrate the data, transform the processes, and deliver a return on investment?
This is the backdrop behind the rapid rise in FDE demand.
According to Christian & Timbers research obtained by TechCrunch, at the beginning of this year only 5% to 10% of surveyed companies planned to hire FDEs, and mostly just for small pilots. By the end of Q2, that figure had risen to 70%.
Some large consulting and services firms have even said they plan to increase their FDE headcount tenfold, forming teams of 20 to 100 people.
The research compiled interviews and surveys conducted between January and June 2026, covering more than 250 senior recruiting executives from 180 companies, 80 executives from Fortune 500 companies, and over 300 FDEs and Applied AI Engineers.
Christian & Timbers founder Jeff Christian said companies are hiring faster than anything he has seen in his past experience, with many firms continuing to expand headcount even during the summer, when recruiting activity is usually slower.
What the market is competing for is not just talent who can help employees install or use AI tools. Chris Taylor, CEO of Ode with Anthropic, which focuses on providing FDE services, said that while many FDEs are capable of helping enterprises deploy Claude Code, very few can build flagship AI product features for a company.
Christian said that currently only a handful of top FDEs can deliver tens of millions of dollars in returns for enterprises—for example, by accelerating lead development and revenue growth, or dramatically reducing the manpower needed for financial analysis and document processing.
AWS invests $1 billion, dispatching thousands of engineers to customer sites
AI companies and cloud providers have also begun expanding their own FDE organizations.
In June, AWS announced an investment of $1 billion to establish a global Forward Deployed Engineering organization, planning to dispatch thousands of engineers to work directly with customers to jointly develop and deploy agentic AI systems.
AWS said that FDEs will work with customers' engineering, operations, and security teams to handle data, governance policies, and existing workflows, shortening AI projects that might otherwise take months to just days.
The goal of such services also includes enabling customers to acquire ongoing operational capabilities. After completing a project, engineers leave behind architecture documentation, operation manuals, knowledge graphs, and trained internal teams, reducing enterprises' long-term dependence on AWS personnel.
For AWS, FDEs can further drive enterprise adoption of its own cloud, models, and agentic AI services. AWS has a large base of existing enterprise customers and infrastructure, and can also identify on-site problems that keep AI projects from reaching production.
OpenAI establishes a Deployment Company, acquiring about 150 deployment specialists
OpenAI, meanwhile, established OpenAI Deployment Company in May to help enterprises bring AI systems into daily operations. OpenAI also agreed to acquire Tomoro, an applied AI consulting and engineering firm. Once the deal closes, the new company is expected to start with about 150 Forward Deployed Engineers and Deployment Strategists.
According to OpenAI's explanation, FDEs first tackle the specific problems customers face, verify whether AI systems can deliver real results, and then identify reusable patterns across different projects.
This approach lets OpenAI achieve two outcomes at once: helping enterprises deploy models, and bringing recurring needs observed at customer sites back to its product and research teams, where they are further turned into tools, evals, and platform features.
Anthropic targets mid-sized companies lacking internal AI teams
Anthropic has also partnered with institutions including Blackstone, Hellman & Friedman, and Goldman Sachs to establish an enterprise AI services company, serving community banks, mid-sized manufacturers, and regional healthcare organizations.
Anthropic noted that such companies may want to use frontier AI but lack the manpower needed to develop and operate systems themselves. The service team works with customer employees to review workflows, then develops tools for the time-consuming or error-prone steps.
Take a healthcare organization as an example: engineers might work with clinical staff and IT departments on medical record documentation, medical coding, prior authorization, and compliance review. Industry personnel explain the actual working context, while engineers translate those requirements into AI tools that can connect to existing systems.
AWS, OpenAI, and Anthropic each take different organizational and collaboration models, but their direction is similar: sending people with engineering capabilities deep into customer organizations to shorten the distance for AI to move from experimentation into formal operations.
How do FDEs create ROI on AI investments for enterprises?
When enterprises adopt AI, model capability is usually just one of the conditions. Whether they can obtain the right data, set appropriate permissions, establish evals standards, and get employees to adopt the technology all affect project outcomes.
FDE work usually focuses on several areas: identifying business problems suitable for AI, organizing enterprise data, designing how models integrate with existing systems, building evals mechanisms, and tracking changes in revenue, costs, or processing time after deployment.
For example, a document processing system can successfully summarize a small number of files in a demo; in a production environment, it also has to recognize different formats, handle confidential data, log operation histories, and escalate to human handling when the model makes errors.
Without these designs, even if a model performs well in a test environment, it is hard for it to become a system the enterprise uses every day.
Therefore, the value of FDEs is usually measured by several metrics, including shorter processing time, less manual work, higher product adoption, increased revenue, or faster movement of AI projects from prototype to production.
Should enterprises hire FDEs themselves, or hand it over to AI companies?
Some enterprises choose to hire FDEs directly, hoping to keep data, processes, and industry knowledge in-house.
Christian said that enterprises worry that after handing proprietary business processes to an AI company, the company might later use the related knowledge to enter the same market. Industries such as insurance, fintech, healthcare, and gaming have already begun looking for in-house FDEs.
Building a team in-house helps preserve confidential data and operational experience, and also cultivates long-term operations capabilities. However, enterprises must bear the costs of recruiting, management, and technology updates themselves, and the number of fully capable FDEs is currently quite limited.
Entrusting AWS, OpenAI, Anthropic, or consulting firms allows faster access to models, platforms, and deployment experience. During the collaboration, enterprises may need to open up some processes and data, and they must also confirm whether their internal teams will be able to take over operations after the project is completed.
How enterprises ultimately choose usually depends on data sensitivity, internal engineering capabilities, deployment speed, and how critical the relevant systems are to the company's operations.
Will FDEs be replaced by AI agents?
Demand for FDEs may not maintain its current growth rate permanently.
Christian believes that within the next two years, AI agents may gradually take over some programming, testing, and deployment work. In the medium term, demand for FDEs may also shift toward physical AI, such as helping enterprises bring humanoid robots into production and workflows.
Over the next 5 to 10 years, the FDE role might even gradually disappear.
However, enterprise AI deployment still involves organizational coordination, permission design, risk judgment, and industry regulations. Even if AI agents can complete more technical work, enterprises still need someone to decide which problems to solve, how to measure results, and who is responsible when systems fail.
The work of FDEs may therefore change. The amount of code engineers write themselves may decrease, while the share of system design, performance management, risk control, and industry integration may increase.
For AI companies, another challenge is how to scale the FDE business. With each additional customer, the team may need to re-understand data, permissions, and workflows, and delivery costs rise accordingly.
AWS hopes customers will operate systems themselves after projects end, while OpenAI tries to turn recurring needs across different projects into reusable tools. Whether FDEs can turn one-off custom experiences into product capabilities that subsequent customers can use directly will affect the long-term profitability of this model.
At least at this stage, AWS, OpenAI, and Anthropic successively expanding enterprise deployment services shows that the competitive standards of the AI market are changing. Model capability will still influence enterprise choices, but whether models can be connected into real workflows, run stably, and deliver ROI is also beginning to shape the business prospects of AI companies.
Editor: Sisley Copy Editor: Mia Sources:
This article's first draft was written with AI assistance by INSIDE and reviewed and confirmed by human editors; get featured articles delivered straight to you—click to subscribe to the INSIDE newsletter now!
Further reading:
Latest developments: