Silicon Valley’s Costly Illusion: The Rise of FDEs

By Anurag Gurtu

Congratulations to the enterprise artificial intelligence venture that recently secured another nine-figure funding round to address a challenge that software ought to resolve on its own: persuading humans to actually use the program.

You recognize the type. No names necessary. You have likely perused the promotional press release—featuring buzzwords like “autonomous,” “agentic,” and “category-defining” in the opening paragraph—only to find later down the page where the capital is truly headed: recruiting an army of engineers to sit within client offices and manually force the supposedly autonomous system to function properly.

Nothing screams “we’ve engineered the future of work” quite like acquiring half a billion dollars in funding solely to hire extra people to execute the work.

The business model, decoded

Here is the underlying mechanic, in case the pitch deck was intentionally opaque:

1. Sell the client an “AI platform.”
2. Realize the platform fails to perform the task independently.
3. Dispatch forward-deployed engineers: brilliant, recently graduated computer science degree holders who now spend six months hand-crafting the workflows that the corporate website implied were already automated.
4. Bill the client for “software.”
5. Record the income as Annual Recurring Revenue (ARR).
6. Secure a larger funding round on the premise that “ARR is climbing.”
7. Iterate indefinitely, maintaining a progressively larger and costlier engineering organization positioned squarely between the product and its original promise.

This does not constitute a virtuous business flywheel. It is simply a payroll masked by a slide presentation.

Worse yet, the sector has abandoned even the pretense of concealment—it actively promotes and raises funds based on this exact practice.

OpenAI committed up to $1.5 billion toward a deployment joint venture valued at $10 billion, alongside an additional $4 billion contributed by TPG, Bain Capital, Advent, Brookfield, and Goanna. The resulting entity, DeployCo, has reportedly amassed upwards of $4 billion in total investments.

Meanwhile, open positions for “forward deployed engineer” roles have surged between 800% and 1,000% year over year. Lightcast figures highlighted by Fortune noted a jump exceeding 1,000% between January and August 2026. Financial Times data cited by PYMNTS showed listings advancing over 800% from January to September 2025, while Indeed figures indicated job openings climbed from 643 in April 2025 to 5,330 in April 2026. Total compensation packages in the United States are scaling past $300,000. Active employer listings place Anthropic’s base pay at $280,000 to $320,000 and OpenAI at $185,000 to $300,000, not including equity grants. Fortune also documented select Anthropic positions reaching total compensation packages of $400,000.

Somewhere out there, a venture capitalist is labeling this arrangement “operating leverage” with a completely deadpan expression.

It is not operating leverage. It is merely a traditional staffing agency that has learned to adopt the word “agentic.”

The tell

Pose the singular question that every pitch deck purposefully avoids: “If you pulled the embedded engineer out tomorrow, does the client still receive the utility they paid for?”

If the truthful response is “well, no, but the ARR is genuine”—that does not qualify as product-market fit. That is people-market fit, and people-market fit possesses a harshly restrictive ceiling: You can only recruit human labor at a finite speed, and you can never hire them cheaply enough to justify a software valuation multiple. Every additional new account does not automatically become more profitable to service. Instead, it triggers yet another headcount requisition.

Consider two enterprises, each generating $20 million in ARR. One enterprise requires 10 engineers to maintain operations. The other demands 100. Both slide decks declare them to be “AI software companies.” The actual divergence manifests directly within gross margins, the cost required to onboard subsequent clients, and whether that overhead decreases as the organization scales.

The bill always comes due

Here is the reality regarding people-market fit: it feels phenomenal right up until someone poses inconvenient inquiries: What is the fully loaded cost required to onboard the subsequent customer? Does that metric decline as the business expands, or does it simply inflate the organizational chart? What happens to gross margins if initial growth decelerates and headcounts can no longer hide behind hypergrowth metrics?

Businesses backing their forward-deployed engineering armies with newly secured nine-figure funding rounds are banking on the assumption that nobody will raise these questions until the round closes and the celebrations conclude. That is a safe gamble during a hype cycle, but a disastrous strategy in a marketplace that is beginning to parse the fine print.

Because eventually, someone always reads the fine print. A chief financial officer constructs a cost-per-deployment model. A chief information officer benchmarks time-to-value across three distinct vendors rather than trusting a single promotional demo. An investor finally executes the tedious task of separating authentic software ARR from human-subsidized ARR on the exact same balance sheet.

The core distinction buyers and investors can evaluate lies directly in the metrics. Traditional software ARR should scale with declining delivery costs. Human-powered ARR does not share this trait. Notably, OpenAI guaranteed the private-equity backers of its deployment vehicle a fixed annual return of 17.5% over five years, which represents a financial structure entirely distinct from a standard software growth thesis.¹

So no, your artificial intelligence platform is not genuinely autonomous. Your business model, however, just might be—autonomously shifting the financial burden of a services firm onto a software valuation for precisely as long as the music keeps playing.

Enjoy the financing round. Hopefully, the bloated organization chart proves worthwhile.

---

Anurag Gurtu, the co-founder and CEO of Airrived, is an entrepreneur and product leader with more than 20 years of experience across cybersecurity and AI. He has built three generative AI companies, including one that reached unicorn status. Gurtu is known for turning advanced AI into scalable, real-world platforms, and previously contributed to successful exits, including acquisitions by Splunk, Tripwire and Sun Microsystems, and a public company, FireEye. He holds a master’s degree from the University of Southern California and CISSP certification.

Illustration: Dom Guzman

Frequently Asked Questions

What is a forward-deployed engineer (FDE)?

In the context of enterprise AI, FDEs are technical personnel—often holding computer science degrees—sent directly into customer offices to manually configure, build workflows, and troubleshoot software solutions to ensure they function as promised.

Why is human-powered ARR criticized compared to software ARR?

Software ARR scales efficiently with decreasing delivery costs over time. Human-powered ARR requires constant headcount growth for every new customer acquired, limiting profitability and failing to support traditional software valuation multiples.

How much have job postings for FDEs increased?

Data from various sources including Lightcast, Financial Times, and Indeed show FDE job postings growing between 800% and 1,000% year-over-year during 2025 and 2026, with total compensation packages frequently surpassing $300,000 in the U.S.

What financial guarantee did OpenAI provide to its deployment vehicle backers?

OpenAI guaranteed the private-equity backers of its deployment vehicle an annual 17.5% return over a five-year period.

---