The FDE Bubble: Why Enterprise AI's Hottest Role May Signal a Deeper Problem

Quick Facts

• Job postings for forward deployed engineers grew roughly 729% between April 2025 and April 2026, according to Indeed data.
• OpenAI launched a deployment joint venture valued at $14 billion in May 2026; Anthropic followed with its own PE-backed venture valued at $1.5 billion.
• Gartner expects more than 40% of agentic AI projects to be cancelled before the end of 2027.

Enterprise AI companies are building armies of engineers who embed inside customer organizations to hand-build what their sales decks say already exists. The practice is called forward deployed engineering, and it is now the dominant delivery model for enterprise AI. That may be the industry's biggest problem.

Anurag Gurtu, co-founder and CEO of Airrived, published the argument directly. His company was named among 11 startups globally in Gartner's agentic AI vendor report, out of 129 companies evaluated. He says the FDE model hides a staffing business inside a software valuation.

What Is a Forward Deployed Engineer?

The role originated at Palantir roughly two decades ago. Engineers embedded inside government clients, working on secure, air-gapped networks to build systems on live data. The model worked because those deployments were genuinely one-of-a-kind.

The role spread across the AI industry in 2026 once companies discovered the hard part of enterprise AI is deployment, not the model. A forward deployed engineer embeds directly inside a customer's organization to build, integrate, and ship working systems on that customer's own infrastructure.

The Numbers Behind the Boom

Indeed tracked roughly 729% growth in FDE job postings between April 2025 and April 2026. Separate survey data shows the share of companies planning to hire FDEs for AI went from 5-10% to above 70% among 180 companies polled. Gartner expects 85% of technology providers to have launched forward-deployed engineering programs by the end of 2026.

The pay reflects the demand. Median mid-level FDE total compensation in 2026 is $385,000. Staff-level engineers clear $610,000. Principal FDEs at frontier labs reach $1.2 million. Andrew Ng noted on X that FDEs are among the buzziest new roles in Silicon Valley, embedded inside clients to build and tune agentic workflows for specific organizational needs.

The Business Model Problem

Gurtu frames the financial flaw with a direct comparison. Two companies, both at $20 million in ARR. One needs 10 engineers to operate. The other needs 100. Both pitch themselves as AI software companies. The difference appears in gross margin and in whether the cost to onboard the next customer falls over time.

Software economics improve with scale. Human-powered delivery does not. Gurtu calls this the tell that separates a software business from a staffing operation wearing a software valuation.

He also flags a lock-in risk buyers often miss. Deployment accounts for roughly 20% of an AI system's lifetime cost. The remaining 80% is operation and ongoing change. When an FDE ships a working system but leaves no documentation or internal capability behind, the customer has not bought a solution. The dependency and its cost remain.

Big Labs Are Doubling Down Anyway

OpenAI launched what it calls the Deployment Company on May 11, 2026. The joint venture carries a post-money valuation of $14 billion. TPG leads with $4 billion in committed capital, with Advent, Bain Capital, and Brookfield as co-leads. McKinsey, Bain and Company, and Capgemini joined as founding partners on the consulting side. OpenAI also acquired Tomoro, an applied AI engineering firm, bringing roughly 150 experienced FDEs into the new entity from day one.

Anthropic moved in the same direction. Blackstone, Hellman and Friedman, and Goldman Sachs anchored a new enterprise deployment joint venture valued at $1.5 billion, with $300 million committed from each of the three lead partners. Apollo Global Management, General Atlantic, and Sequoia Capital joined as additional backers.

Gurtu reads the private equity interest clearly. A guaranteed return structure on a deployment arrangement is not a growth thesis. It is a fixed-income trade structured around labor, not software.

What Buyers and Investors Should Watch

MIT research published in 2025 found roughly 95% of enterprise generative AI pilots produced no measurable return. Against that backdrop, the FDE model looks like a response to a real problem: AI that works in demos does not work in production without significant human effort.

The question for buyers is whether that effort ever ends. FDE-as-a-service arrangements are now being packaged and sold as ongoing engagements rather than one-time deployments. For investors, the margin question is equally direct: if headcount must grow with revenue, the software multiple does not apply.

Of $111 billion in scale-up investments in Silicon Valley in 2025, 93% went to AI companies. Whether that capital lands on software businesses or expensive services operations may depend on which questions buyers and investors choose to ask.