OpenAI Bought 150 FDEs. The Real Pool Is 5,100 SAs at Bain and Capgemini.

Three frontier labs stood up forward-deployed engineering arms inside 48 hours. Google Cloud listed 59 FDE roles on May 12. OpenAI launched DeployCo the same day with $4B and 150 engineers from Tomoro. Anthropic, Blackstone, Hellman & Friedman, and Goldman had announced their own $1.5B JV a week earlier. If you are still typing "forward deployed engineer" into LinkedIn search, you are competing with every other recruiter in tech for the same few hundred profiles. The supply that actually moves the needle is hiding under a different title at firms that just invested in DeployCo.

What actually happened the week of May 12

Google Cloud CEO Thomas Kurian posted on LinkedIn that the company had opened 59 distinct forward-deployed engineering roles across the US, India, Brazil, Australia, Mexico, Singapore, South Korea, and Canada. Seniority spans FDE II through FDE IV. Most listings require RAG architectures, vector databases, foundation model fine-tuning, and production deployment on cloud platforms. US base bands run $127,000 to $183,000 for Applied FDE up to $183,000 to $265,000 for FDE IV, before bonus and equity.

On the same day, OpenAI announced the OpenAI Deployment Company, a majority-owned subsidiary capitalized with more than $4 billion in initial funding from a TPG-led syndicate. Advent International, Bain Capital, and Brookfield are co-lead founding partners. Fifteen other investors include SoftBank, Goldman Sachs, Warburg Pincus, and three consultancies whose presence is the actual story: Bain & Company, Capgemini, and McKinsey. DeployCo investors are guaranteed a 17.5% minimum annual return over five years with profits capped. The 19 partners collectively sponsor more than 2,000 portfolio companies, pre-sold into the pipeline.

The acquisition wrapper for the launch was Tomoro, a London-headquartered AI consultancy founded in 2023 by Rishabh Sagar, Albert Phelps, Chris Spencer, Ed Broussard, Chloe Kelleher, Ash Garner, and Sandi Chanda. Tomoro's roughly 150 AI engineers and deployment specialists are now DeployCo's seed bench. Their client list (Tesco, Virgin Atlantic, Supercell, Mattel, Red Bull, Fidelity International) tells you the segment.

Eight days earlier, Anthropic, Blackstone, Hellman & Friedman, and Goldman Sachs unveiled a competing $1.5B JV with Apollo, General Atlantic, Leonard Green, GIC, and Sequoia along for the ride. Blackstone President Jon Gray called the scarcity of engineers who can implement frontier AI "one of the most significant bottlenecks to enterprise AI adoption." He is correct, and that is precisely the problem for anyone running a search this quarter.

The supply ceiling is lower than people think

5,100

Refolk's index shows roughly 5,100 solutions architects worldwide whose profiles include generative AI, LLM, or ML skills. That is the entire title-matchable pool on Earth, before you account for who is willing to move, who is on a non-compete, who lives in a region you can hire in, and who is locked up by an equity refresh.

Now layer in the demand side. Google Cloud wants 59 immediately. OpenAI absorbed 150 from Tomoro and is funded to acquire more. Anthropic's JV needs to staff a parallel firm from scratch. Salesforce has publicly committed to 1,000 FDEs. The math does not work without poaching, and the only places with enough qualified solutions architects to poach from are the same SI partners now sitting on DeployCo's cap table.

Filtering on the literal title "FDE" misses 95% of the people who can actually do the job. </pull>

Why "FDE" is title arbitrage, not a profession

Palantir invented the modern FDE in 2011 as a combined solutions-and-integration engineer. Andreessen Horowitz later called the recasting "title arbitrage." That framing matters for sourcing. The behavior pattern that makes an FDE is not in the title field. It is: customer-facing, ships production code, owns outcome metrics, comfortable in front of a CIO on Tuesday and a Postgres console on Wednesday.

Filter on the title and you find the people who already negotiated for the title. Filter on the behavior and you find five times the pool, mostly under labels like:

• Solutions Architect (the volume play)
• Delivery Principal / Delivery Lead
• Engagement Engineer / Engagement Manager (technical track)
• Customer Engineer (Google's internal label)
• ML Engineer, Customer-Facing
• Applied AI Engineer
• Specialist Solutions Architect

This is the core problem in forward deployed engineer sourcing right now. The supply you actually want is hiding two synonyms away from your Boolean. This is the kind of pattern matching Refolk was built to do: describe the behavior in plain English ("customer-facing engineer who ships RAG pipelines into Fortune 500 banks, currently at a Big Four or Tier 1 SI") and the system pulls a ranked shortlist across LinkedIn, GitHub, and the open web, regardless of what the title field says.

The investor list is the hunting list

Read the DeployCo press release like a sourcer, not like a journalist. Three of the 19 investors are management consultancies (Bain & Company, Capgemini, McKinsey) whose enterprise integration revenue this venture most directly threatens. That investment is a hedge, not a strategic alignment. They put money in to keep a seat at the table while OpenAI eats their lunch.

Two consequences flow from that. First, their senior solutions architects with GenAI experience are the exact people DeployCo, Google Cloud, and the Anthropic JV need to staff against. Second, those firms now have an explicit financial interest in retaining them. Expect counter-offers to escalate sharply through Q3 2026. Budget for it in your offer model now.

The named feedstock for any solutions architect to FDE pipeline this year:

• Big Four / GSI bench: Accenture, Deloitte, Capgemini, McKinsey QuantumBlack, BCG X, EY. These were already part of OpenAI's February 2026 Frontier Alliances program, which is the prequel to DeployCo and explains how OpenAI knew which firms to invite onto the cap table.
• Hyperscaler SA orgs: AWS, Microsoft, Google Cloud (cannibalizing itself), Oracle's OCI specialist track.
• Regional SIs for the mid-market track: Slalom, West Monroe, Publicis Sapient, Thoughtworks. This is where the Anthropic JV will land most of its hires, because Goldman's Marc Nachmann explicitly positioned that venture toward mid-market deployment.
• Boutique AI consultancies: Tomoro was the template. Expect 3 to 5 more 50 to 200-person acqui-hires in the next 12 months. Founders of those shops are also worth tracking directly.

Geography: where the Tomoro 150 actually sit

Tomoro is headquartered in London with offices in Edinburgh and Manchester, plus an APAC HQ in Singapore and presence in Sydney and Melbourne. Those are the cities where DeployCo will be staffing densely in 2026, both because the existing bench is there and because the SI partner offices that fund it are co-located.

For OpenAI Deployment Company hiring, the four cities to watch hardest are London, Singapore, New York, and San Francisco. For Google Cloud FDE jobs, add Bangalore, Mexico City, and Seoul. For the Anthropic JV, expect a heavier US mid-market footprint (Chicago, Atlanta, Denver) because that is where the target customer base actually buys.

If you are hiring against any of these, geo-fenced searches by skill graph beat title search by a wide margin. The Anthropic JV will not be posting a job called "Anthropic Forward Deployed Engineer" in Atlanta in June. They will be calling a partner at Slalom and asking who the strongest customer-facing ML engineer on the bench is. Your job as a recruiter is to know that name before the partner does.

The two-track sourcing strategy for FDE recruiting 2026

OpenAI is going Fortune 500. Anthropic is going mid-market. That splits the talent profile cleanly and you should run two separate searches.

Track 1: Fortune 500 (OpenAI DeployCo, Google Cloud, Salesforce)

Look for SAs with 7+ years at a Big Four or GSI, named-account experience with banks or retailers, evidence of production GenAI work (conference talks, internal case studies, partner certifications at the Specialty level), and CIO-facing comfort. Compensation expectations are already set against the published Google FDE bands, so anchor offers above $200K base for senior profiles.

Track 2: Mid-market (Anthropic JV, regional rollouts)

Look for SAs with 4 to 8 years at a regional SI or a hyperscaler's commercial segment, generalist breadth, vertical depth in one or two industries, and a portfolio of 8 to 15 smaller deployments rather than 2 to 3 mega-projects. Compensation anchors lower but equity in the new JV may be more meaningful.

Either way, the first filter is skill evidence, not title. We use Refolk internally for exactly this: a query like "engineers who have presented a RAG production case study at a vendor conference in the last 12 months and currently work at a top-20 SI" returns a different set of people than any LinkedIn title search will, and it is the set that actually answers the brief.

The 90-day window

Here is the practical timeline. DeployCo announced May 12. Tomoro integration closes through Q2. First Frontier Alliance customer engagements that need staffing beyond the 150 transferred engineers begin in Q3. That means the named SI partners will be quietly pulling their best GenAI-capable SAs onto DeployCo loaner arrangements through July and August, and those people will not be on the market in the conventional sense. They will be on the market for the right inbound message.

If you are a founder or eng leader competing against DeployCo for the same profiles, you have roughly 90 days before this pool gets meaningfully tighter and another 180 before counter-offer inflation reprices the band. Build your shortlist now, against the behavior pattern, across the 25 or so firms named above. The title search will not find them. The skill search will.

FAQ

Is "forward deployed engineer" a real job category or a marketing label?

Both, depending on the firm. Palantir formalized the role in 2011 and a few companies (OpenAI, Anthropic, Google Cloud, Salesforce) now use the title operationally. At most other firms it is title arbitrage applied to what used to be a senior solutions architect or delivery principal. For sourcing purposes, treat it as a behavior pattern (customer-facing, ships production code, owns outcomes) and ignore the title field.

Where should I look first for FDE candidates in 2026?

Start at the named SI partners who invested in or partnered with the labs: Bain & Company, Capgemini, McKinsey, Accenture, BCG, Deloitte. Then hyperscaler SA orgs at AWS, Microsoft, and Google Cloud (which is now cannibalizing itself for its own FDE roles). For mid-market roles, add Slalom, West Monroe, Publicis Sapient, and Thoughtworks. Filter on GenAI skill evidence inside those firms rather than on title.

How does the OpenAI Deployment Company differ from Anthropic's JV for sourcing?

OpenAI's DeployCo is capitalized at $4B and aimed at Fortune 500 deployments through its 19 SI and PE partners' 2,000-plus portfolio companies. Anthropic's $1.5B JV with Blackstone, Hellman & Friedman, and Goldman is positioned toward mid-market. The profile splits accordingly: senior Big Four / GSI talent for OpenAI, generalist regional-SI talent for Anthropic. Run them as two separate searches.

What about counter-offers?

Plan for aggressive counters from the SI partners that just invested in DeployCo. Those firms have a financial interest in retaining the same SAs you are recruiting. Bring a best-and-final offer to the first conversation when possible, and lean on equity in newer ventures (the Anthropic JV, smaller boutiques being rolled up) where the counter-offer math is harder for incumbents to match.