Neocloud capital formation continued at pace through Q1 2026. We tracked disclosed raises across roughly 30 companies in the GPU-as-a-service and AI-infrastructure category, totaling over $18B in announced commitments. The deals are getting larger, the structures are getting more sophisticated, and the use of proceeds is shifting away from pure GPU procurement toward power, real estate, and team build.
Q1 totals at a glance
That total includes equity rounds, structured debt facilities, and announced strategic commitments where the deployment timing is disclosed. It excludes private credit facilities where the size isn't public, which would push the true number higher.
Notable equity rounds
- $1.5B Series D-class round for a top-five neocloud at $14B post-money. Lead: large public-market crossover investor with a sleeve into private growth.
- $850M Series C for a sovereign-adjacent AI compute provider focused on EMEA. Strategic backing from a national investment fund.
- $600M Series B for an emerging neocloud specialized in inference workloads, valuation undisclosed but estimated at 6-8B post.
- $420M Series B for a vertically-integrated power-plus-compute play, with a meaningful sleeve earmarked for site acquisition.
- Multiple sub-$200M Series A rounds across roughly a dozen earlier-stage companies, suggesting continued breadth in the category despite consolidation pressure at the top.
Debt and structured facilities
The bigger story is on the debt side. Multiple multi-billion-dollar structured facilities closed in Q1, backed primarily by GPU collateral with secondary recourse to operating cash flow. The pricing is tightening as the asset class matures: senior secured pricing that was SOFR+650-800 in 2024 is now closer to SOFR+450-600 for established operators.
- Three disclosed facilities of $2B+ each at major neoclouds, primarily GPU-backed.
- Several $500M-$1B facilities at mid-size operators, mixed collateral.
- First disclosed CMBS-style facility backed by data center real estate at a top-tier neocloud, suggesting the financing toolkit is broadening.
- Sovereign and quasi-sovereign capital appeared in more deals than in any prior quarter — a notable shift in the LP composition of the sector's capital stack.
Where the money is being deployed
Use of proceeds is shifting away from the 2023-2024 default of 'GPU procurement + standard colo lease' toward more vertically integrated structures. Three patterns:
- 01Site acquisition or long-dated leases — operators are buying or leasing capacity directly, rather than relying on third-party colo. We see meaningful sleeve allocations specifically for site control.
- 02Power procurement and behind-the-meter generation — the operators that raised this quarter are explicitly funding power infrastructure, not just IT load. Several disclosed PPAs and gen-tie investments.
- 03Geographic expansion outside the US — Europe, Middle East, and APAC capacity announcements outnumber US announcements for the first time in any quarter we've tracked.
What the data tells us
First, the category is not slowing. Whatever the right framework for AI demand is, the capital flowing into the underlying infrastructure assumes durable, long-horizon growth. Second, the survivors are getting larger; the gap between the top five and the next twenty is widening. Third, vertical integration is now the default thesis — operators that can't control power, real estate, and operations end-to-end are losing rounds to operators that can.
“$18B in a quarter into the same broad asset class — at increasingly sophisticated structures — is a clear signal. The infrastructure layer of the AI cycle is still in expansion mode.”
What we expect through Q2
Three things we'd bet on. Continued capital formation at scale; the round sizes don't compress in the near term. Continued differentiation by operator type — pure GPU rental, vertically integrated, sovereign-adjacent. And the start of more aggressive M&A as the gap between haves and have-nots widens; we expect at least two notable acquisitions or roll-ups by mid-year.
