Matchr Research · 2026

    The State of AI Infrastructure Hiring 2026

    Capital is committed. Compute is being built. The people are the bottleneck. A fully sourced reference for CHROs, VPs of Talent, and Heads of Recruitment planning AI infrastructure headcount against capex that will not wait.

    Created byThe Global Embedded RPO Company

    Get the report

    The shape of the problem

    The numbers your hiring plan is up against

    $318B → $1T+

    AI infrastructure spend, 2025 → 2029

    46%

    of operators report difficulty finding qualified candidates

    ~1%

    of mechanical & electrical engineers globally have data center experience

    $14.7M

    projected agency savings at Nebius (case study)

    The capex side of the AI boom is well covered. The talent side is not.

    Every analyst report and earnings call talks about GPUs, megawatts, and capex. Very few talk about the people actually needed to build, operate, and run the infrastructure — and what it takes to hire them at the speed the business is moving.

    We sit inside this market every day. Across our embedded RPO engagements, we see hiring plans approaching 1,000 roles in 12 months become the norm for the neocloud cohort. We see talent functions of 10-15 people asked to deliver what historically took a team of 50 over three years. We see senior offers lost to AWS, Google, and Meta on package shape, not headline number.

    This report puts the public data side by side with what we observe operating inside these engagements, so AI infrastructure talent leaders have a single, defensible reference for what they're up against. It is built from IDC, McKinsey, Uptime Institute, Deloitte, BLS, CBRE, JLL, SEC filings, and our own Nebius engagement, with sources cited throughout.

    If you are planning hiring for an AI infrastructure scale-up in 2026, this is for you.

    “Capital is committed. Compute is being built. The next constraint is people. The work begins now.”
    Adriaan Kolff

    Adriaan Kolff

    CEO & Co-founder, Matchr

    What's inside

    Built to survive scrutiny in a board meeting

    Eight sections that move from the public record to the operating reality: what is committed, where the talent actually sits, what it costs to compete for it, and what the strongest talent functions are doing about it. Every claim cited to IDC, McKinsey, Uptime Institute, Deloitte, CBRE, JLL, BLS, SEC filings, or our own embedded engagement at Nebius.

    • The compute boom is no longer hypothetical

      IDC, McKinsey, and CoreWeave/Nebius/Crusoe filings — what's actually been committed, built, and forecast through 2029.

    • The talent pool is structurally constrained

      Uptime Institute hiring data, Deloitte posting growth, BLS engineer counts. Why "skilled-labor shortage" beats every other obstacle in executive surveys.

    • Where the people actually are

      Top 13 global data center metros by installed capacity. Why a senior requisition outside the clusters becomes a 9-to-12 month problem, and what to do about it.

    • The compensation reality

      Levels.fyi data on hyperscaler hardware engineering comp — why the scale-ups winning senior hires from AWS are rarely the ones offering the largest cash package.

    • The neocloud scaling paradox

      CoreWeave, Nebius, Lambda, Together AI, Crusoe — capital raised, capex committed, headcount doubled in 12 months. Public filings, all sourced.

    • How Matchr is helping Nebius scale at AI-infrastructure speed

      A real engagement, in concrete numbers. $1.3M agency spend already displaced, 80% of agency motion eliminated, first hires landed in month one. Full case study inside.

    A preview

    The charts your board will ask about

    IDC spending forecasts, Uptime and Deloitte hiring data, global metro capacity from CBRE and JLL, Levels.fyi compensation benchmarks, and the five-neocloud capital and headcount table. Click any chart to open it.

    Figure 01 / 05

    AI infrastructure spending more than doubled in 12 months. Forecast above $1 trillion by 2029. Source: IDC, April 2026.

    Who this is for

    Built for talent leaders at AI infrastructure scale-ups

    If you own hiring at a neocloud operator, an AI infrastructure company, or any hyperscaler-adjacent business competing for the same specialist pool, this report is built for you.

    01

    CHROs and CPOs

    You are planning headcount against capex that is moving faster than you can hire.

    • A defensible reference for board and CFO conversations on talent capacity
    • Sourced benchmarks on hiring difficulty and skills gaps
    • The hyperscaler comp gap — and what closing it actually requires
    02

    VPs of Talent · Heads of Recruitment

    You are running a TA team against a hiring roadmap the size of a Series C.

    • Where the candidate pools actually live, with quantified geography data
    • Internal-build vs embedded-vs-agency math, applied to a real 361-hire plan
    • Time-to-productivity benchmarks for specialist recruiters
    03

    CFOs and capital allocators

    You are evaluating whether the talent function can absorb the capex you have committed.

    • The agency cost curve at scale — $16.6M baseline math, sourced
    • Cost-per-hire benchmarks against the embedded RPO alternative
    • What capex deployment risk looks like when hiring is the bottleneck
    “Scaling internally would have been too slow, and leaning on agencies would have been expensive and fragmented. Matchr offered something different: experienced recruiters embedded in our team, working as an extension of the business. We'll be close to 3,000 employees by end of 2026, and Matchr has been a core part of getting us there.”
    Marcus Pask

    Marcus Pask

    Talent Acquisition Leader @ Nebius, Nebius

    Newsletter

    Get the monthly Matchr update

    A monthly email with our latest insights, podcasts, upcoming events, and new opportunities in talent acquisition.

    Do you need help hiring for your AI infrastructure company?