Case study
Nebius entered 2026 with a 310-megawatt AI data center under construction, a $16–20B capex plan, and a hiring roadmap of 361 senior hires. Matchr embedded 13 senior talent partners across four business lines and saved $1.3M on agency spend in the first four months.
How Matchr helps Nebius scale at neocloud speed
The impact, in numbers
Saved on agency spend in the first 4 months
Agency spend eliminated on the roles Matchr works on
Projected net savings by end of 2026
Industry
AI Infrastructure
Service
Embedded RPO
Duration
Ongoing since January 2026
Roles hired
Data center operations and engineering, Hardware engineering, Network engineering +3
The Challenge
By the start of 2026, Nebius was deploying compute at neocloud-leading pace. A 310-megawatt AI data center under construction in Finland. Full-year 2026 revenue guidance of $3.0 to $3.4 billion, up from $530 million in 2025. A 2026 capital expenditure plan of $16 to $20 billion.
The talent function was running against the same curve. Nebius was around 700 people with a twelve-person internal TA team, heading into a major growth phase that would take headcount close to 3,000 by the end of 2026. The hiring roadmap approached 361 senior hires across data center, GTM, R&D, and corporate functions in the year ahead.
Sourcing was inbound-led in a market where the candidate pool is structurally narrow. Agency invoices were running at $46,000 per senior placement.
Nebius came to us at the start of a crazy scaling journey. They told us up front: we don't have the capability or the bandwidth to do this ourselves. Three months in, the embedded team is thirteen senior partners across four business lines, and the first offers were going out inside a month.
Elliot Read, Managing Director at Matchr
The Solution
Why Nebius chose embedded RPO
Two paths sat in front of Nebius before the embedded RPO conversation began. The first was building specialist talent acquisition capacity from scratch, hiring and training senior recruiters internally. The second was layering on more staffing agencies. The first was too slow against a 361-hire roadmap, with each senior recruiter taking around six months to reach productivity. The second was too expensive, with senior placements priced at $46,000 each, equivalent to a $16.6M agency baseline for the year if every hire ran through external recruiters.
Cost was part of the case for embedded RPO. It was not the deciding factor. Embedded means dedicated: an embedded talent partner works only on Nebius, while an agency recruiter splits time across many clients. Embedded means partnership rather than transaction: hiring managers and recruiters share the same Slack channels, the same incentives, and the same accountability. Embedded means continuity: the recruiter who sourced a senior data center engineer six weeks ago is the one closing the next senior data center engineer today, with full context on what worked. Embedded means outbound by default, in a market where inbound-only sourcing leaves the rarest candidates untouched.
For Nebius, those four properties — dedicated focus, partnership posture, continuity of context, outbound-led sourcing — were the reasons embedded RPO won the decision. The cost saving is the consequence, not the cause.
What Matchr brought to Nebius
Three constraints we see most often across our AI infrastructure book, and how each was addressed inside the engagement.
1. Cost efficiency
The constraint. Nebius was running heavy agency reliance, with senior placements priced at $46,000 each and a 361-hire plan that would have produced a $16.6M agency baseline for the year.
What we did. Replaced the inbound, agency-led motion with outbound-led, dedicated sourcing across all four business lines.
Result. $1.3M saved year-to-date on agency spend across the roles Matchr works on, with 80% of agency spend eliminated on those roles.
2. Ramp speed
The constraint. A specialist internal hire takes around six months to reach productivity in this market. Funding rounds and capex deployment timelines do not wait that long.
What we did. Deployed senior recruiters who started inside a week, with no training required and full domain context on day one.
Result. First hires landed in month one, not month six. Nebius hiring at full speed while competitors were still building their TA teams.
3. AI specialists
The constraint. Only roughly 1% of mechanical and electrical engineers globally have data center ops experience. Generalist recruiters cannot credibly source, screen, or close inside that pool.
What we did. Embedded 13 senior specialist talent partners across data center, GTM, R&D, and corporate functions to hire from the top 1%.
Result. Recruiter depth matched to candidate depth on every desk, with senior partners productive in week one.
The Results
The headline result, four months into the engagement, is $1.3 million in net agency savings across the roles Matchr works on. Net monthly savings ramped from approximately $2,000 in January (an onboarding month with the embedded team standing up) to $211,000 in February, $384,000 in March, and $752,000 in April.
The trajectory reflects the team reaching steady-state delivery: thirteen senior partners productive in week one, sourcing motion fully outbound by month two, and agency placements displaced rather than supplemented.
Projected to the end of 2026, the engagement displaces $14.7 million in net agency spend against a 361-hire roadmap. The arithmetic is direct: 361 hires at the $46,000 agency baseline is a $16.6 million annual cost, against which Matchr's 12-month embedded RPO fee of approximately $1.9 million represents an order-of-magnitude reduction.
The takeaway
Four months in, the engagement has already moved the cost curve, the speed curve, and the specialist-access curve at the same time. $1.3 million in agency spend has been displaced, with 80% of the agency motion eliminated on the roles Matchr works on. First hires landed inside month one rather than month six. Thirteen senior specialist partners are operating at full tempo across data center, GTM, R&D, and corporate functions. The projected full-year picture is $14.7 million in net savings against the agency baseline, on a hiring plan that does not slow.
Source
Matchr embedded engagement at Nebius. January to April 2026 are realized actuals. May to December are projected at steady state ($1.7M/month). Matchr fixed monthly cost reaches $168K from March. Agency baseline calculated at $46,000 per senior placement.
Appendix · Additional notes
Further context
Why this matters for AI infrastructure talent leaders
Reading the public data and the Nebius engagement together, three implications follow for any CHRO, VP of Talent, or Head of Recruitment at an AI infrastructure scale-up planning the next 24 months.
1. The embedded model breaks the agency cost curve
When monthly hiring targets cross into double digits per sub-function and per-placement fees average $46,000 on senior technical roles, agencies become a financial liability rather than a delivery partner. A 361-hire annual plan at the agency baseline is a $16.6M cost. The embedded RPO equivalent runs at roughly $1.9M for the year, a difference of $14.7 million in net displaced spend.
2. The embedded model collapses the time-to-productivity gap
Building a specialist internal talent function from scratch (including hiring senior recruiters with data center experience, training them on internal systems, and bringing them to productivity) takes around six months per senior hire. The Nebius engagement reached first hires inside month one rather than month six, with the embedded team productive in week one.
3. Only the top 1% of engineers can fill these roles
Only roughly 1% of mechanical and electrical engineers globally have the operational data center experience AI infrastructure roles demand. Generalist recruiters do not close senior data center, hardware, or AI infrastructure engineers. The companies that build privileged access to the specialist pool, internally or through an embedded partner, win the talent war.
Ready to scale your team like Nebius did?
Let's talk about your hiring goals.