GPU Infrastructure: A 2025–2026 Sector Financing Roundup
AI infrastructure continued to attract significant investment through 2025 and 2026 as demand for advanced GPUs and data center capacity drove substantial capital requirements across the sector.
The period included major hardware releases from NVIDIA alongside a growing number of debt financings for GPU cloud and AI infrastructure providers.
Those transactions suggest that GPU financing is developing into a broader institutional credit market spanning asset-backed lending, delayed-draw term loans, corporate credit facilities, private placements and SPV structures.
GPU Development Continues
NVIDIA introduced its Blackwell Ultra platform in March 2025, extending the Blackwell architecture for AI reasoning, agentic AI and compute-intensive inference workloads.¹
The hardware cycle continued in 2026, with NVIDIA announcing that its Vera Rubin platform was ramping into full production.²
For AI cloud providers and data center operators, each new generation creates additional requirements around GPU procurement, deployment capacity, power and supporting infrastructure. Rapid technology cycles also create an important financing consideration: the physical useful life of GPU infrastructure may extend well beyond the period during which that hardware remains economically competitive for certain workloads.
Recent GPU and AI Infrastructure Financings
Several transactions across 2025 and 2026 illustrate how quickly the financing market has developed.
Crusoe secured a $225 million asset-backed facility led by Upper90 to purchase NVIDIA GPUs and supporting infrastructure.³ In February 2026, Nscale signed a $1.4 billion GPU-backed delayed-draw term loan led by funds managed by PIMCO, Blue Owl and LuminArx to finance contracted GPU deployments across Europe.
CoreWeave followed in March with an $8.5 billion non-recourse delayed-draw term loan secured by high-performance computing infrastructure and an associated customer contract. The transaction received investment-grade ratings and included a floating-rate tranche priced at SOFR +2.25%.⁴
On June 1, IREN closed a $3.65 billion investment-grade GPU financing supporting its Microsoft AI Cloud contract. The financing combined a $1.55 billion DDTL priced at SOFR +2.25% with a $2.10 billion U.S. private placement and was secured by GPUs and associated contracted cash flows.
In July, Nebius announced an approximately $775 million senior secured facility backed by deployed GPU infrastructure and contracted cash flows from an investment-grade customer. The facility was priced at SOFR +2.50%.⁵
Lambda followed in August with a $926 million senior secured term loan B financing GPU infrastructure dedicated to an investment-grade customer. The facility priced at SOFR +3.00% and uses an asset-backed SPV structure with amortization aligned to contracted cash flows and the useful life of the underlying infrastructure.⁶
These transactions span corporate credit facilities, asset-backed loans, delayed-draw term loans, private placements and SPV structures, illustrating how a broader financing ecosystem has developed around large-scale AI compute.
GPU / AI Infrastructure Credit Deals
Borrower | Date | Financing | Structure / Lenders |
CoreWeave | Jul. 2023 | $2.3B | GPU DDTL led by Blackstone / Magnetar; acquisition subsidiary borrower; GPUs supporting customer contracts |
Lambda | Apr. 2024 | $500M | Macquarie-led GPU-backed asset financing; GPUs serving as collateral |
CoreWeave | May 2024 | $7.5B+ | DDTL led by Blackstone and Magnetar; participants included Carlyle, CDPQ, DigitalBridge and BlackRock accounts |
CoreWeave | Oct. 2024 | $650M | Corporate credit facility / liquidity financing from bank syndicate |
Crusoe | Mar. 2025 | $225M | Upper90-led asset-backed facility for NVIDIA GPUs and supporting infrastructure; institutional partners included BCIMC, FS Investments, King Street, Liberty Mutual and ORIX |
Fluidstack | Apr. 2025 | Undisclosed | Macquarie Specialised & Asset Finance GPU financing supporting European AI infrastructure |
Crusoe | Jun. 2025 | $750M | Brookfield Infrastructure Debt credit facility supporting AI infrastructure expansion |
Lambda | Aug. 2025 | $275M | Syndicated senior secured facility led by JPMorgan; subsequently upsized |
Nscale | Feb. 2026 | $1.4B | GPU-backed DDTL led by PIMCO, Blue Owl and LuminArx financing multiple contracted GPU clusters |
CoreWeave | Mar. 2026 | $8.5B | Investment-grade, non-recourse HPC/customer-contract DDTL; floating tranche priced at SOFR +2.25% |
Lambda | May 2026 | $1.0B | Corporate first-lien facility including revolver, term loan and DDTL; proceeds include GPU purchases, working capital and corporate purposes |
IREN | Jun. 2026 | $3.65B | Microsoft-backed GPU financing; $1.55B DDTL plus $2.10B U.S. private placement; secured by GPUs and contracted cash flows |
Nebius | Jul. 2026 | $775M | GPU infrastructure and investment-grade customer cash flows; priced at SOFR +2.50% |
Lambda | Aug. 2026 | $926M | GPU-backed SPV term loan B; investment-grade offtaker; priced at SOFR +3.00%; fully amortizing |
CoreWeave | Aug. 2026 | $2.6B | Dedicated GPU / HPC delayed-draw term loan through financing subsidiary |
Pricing Is Becoming More Differentiated
The recent transactions also provide evidence that GPU credit is becoming increasingly differentiated.
Several investment-grade, contract-backed financings completed in 2026 priced in the SOFR +2.25% to +3.00% range, including transactions for CoreWeave, IREN, Nebius and Lambda.
That does not mean GPU financing broadly prices within that range. Other facilities have priced materially wider depending on the underlying customer, structure and risk profile.
Instead, the pricing dispersion reinforces an important underwriting point: lenders are not simply assigning a value to the GPU hardware. Customer credit quality, contract duration, deployment economics, advance rates, recourse, amortization and residual-value assumptions can materially influence both debt capacity and pricing.
In other words, there is an important distinction between financing GPUs and financing contracted compute infrastructure.
What the Market Is Watching
As these financing structures develop, underwriting is expanding beyond the value of the GPU itself. Lenders increasingly need to assess the customer contract, deployment economics, asset competitiveness and residual value supporting repayment.
Lambda's 2026 financings illustrate the distinction. Its $1 billion senior secured corporate facility announced in May provided capital for GPU fleet and data center expansion, while its $926 million August financing used an asset-backed SPV structure tied to GPUs serving a committed investment-grade customer.
The second transaction was fully amortizing, with repayment aligned to both contracted cash flows and the useful life of the underlying GPU infrastructure.
GPU rental rates, utilization and re-leasing economics are also becoming important indicators as installed capacity grows. Recent industry commentary has highlighted rental pricing as one potential signal of supply, demand and utilization across different generations of GPUs.⁷
KBRA's June 2026 research highlights a related issue: physical useful life and economic life are not necessarily the same.
A GPU may remain fully operational after an initial customer contract expires while becoming less competitive for the workloads that determine its earning power. Workload suitability, hardware competitiveness, power cost per unit of compute, customer renewal, third-party demand and equipment refresh strategy therefore become important to re-leasing assumptions.⁸
The credit analysis also extends beyond the server rack.
Power availability and grid interconnection timing can materially affect when financed infrastructure becomes operational and begins producing revenue. Project readiness, construction execution, power supply and deployment timing can therefore become part of the repayment analysis.⁹
Taken together, lenders financing AI infrastructure increasingly need to consider several interconnected questions:
Who is purchasing the compute, and how strong is the counterparty?
How long is the customer commitment relative to the financing term?
When will the GPUs become operational and begin producing contracted revenue?
How much utilization is required beyond the contracted capacity?
How quickly could newer hardware affect the economic competitiveness of the financed GPUs?
What are the renewal, re-leasing and residual-value assumptions after the initial contract expires?
For companies financing AI infrastructure, debt capacity increasingly depends on more than access to valuable hardware. Contract quality, customer concentration, deployment timing, utilization, power availability and the expected economic life of the equipment can all affect how lenders assess risk and structure repayment.
For venture-backed companies considering debt or structured capital alongside equity, these factors are increasingly important to address before approaching the financing market—not after term sheets arrive.
Exploring financing options for your next stage of growth? Connect with Axis to discuss your capital needs.
About Axis Group Ventures
Axis Group Ventures is a boutique investment banking and strategic advisory firm focused on global debt placement and private-market secondaries for venture- and private equity-backed companies. We work with founders, C-suite executives and investors to develop and execute customized private-market capital solutions. For more information, visit www.axisgroupventures.com.
Sources
NVIDIA, “NVIDIA Blackwell Ultra AI Factory Platform Paves Way for Age of AI Reasoning,” March 18, 2025.
NVIDIA, “NVIDIA Vera Rubin Ramps Into Full Production to Power Agentic AI Factories Worldwide,” May 31, 2026.
Crusoe / Upper90, “Upper90 Closes $225M Credit Facility to Crusoe to Expand AI Cloud Infrastructure,” March 27, 2025.
CoreWeave, “CoreWeave Closes Landmark $8.5 Billion Financing Facility,” March 31, 2026.
Nebius, “Nebius Raises $775 Million in First Secured Debt Financing to Accelerate Global Buildout,” July 17, 2026.
Lambda, “Lambda Closes $926 Million Senior Secured Term Loan B Facility,” August 27, 2026.
William De’Ath, “GPU Rental Rates to Drop, AI Infrastructure to Consolidate,” LinkedIn, September 2026.
KBRA, “AI Compute Financings: Evaluating Re-Leasing Risk for GPUs and TPUs,” June 22, 2026.
KBRA, “Data Centers: Developments and Trends in Project Finance,” January 13, 2026.
Disclosures & Disclaimers
This blog post is provided by Axis Group Ventures for informational and educational purposes only. It does not constitute investment, legal, accounting, or tax advice and should not be relied upon as such. Nothing contained herein should be interpreted as an offer to buy or sell any securities. Any actual offer or solicitation will be made exclusively through formal documentation provided by the relevant issuer or seller.
Axis Group Ventures is not a registered broker-dealer and does not execute, negotiate, or recommend the purchase or sale of securities. Any introductions or private-market support provided by Axis Group Ventures are conducted strictly in an advisory and consulting capacity. Readers should conduct their own due diligence and consult qualified professionals before making any financial decisions.
Investments in private securities involve significant risks, including the potential loss of the entire investment, and are typically illiquid. Past performance does not guarantee future results.
This blog post was developed using Axis Group Ventures’ proprietary market knowledge, professional insights, and years of industry experience, supplemented by large language model technology to assist with drafting and editorial refinement.
DISCLOSURE: Rainmaker Securities, LLC (“RMS”) is a FINRA registered broker-dealer and SIPC member. Find this broker-dealer and its agents on BrokerCheck. Our relationship summary can be found on the RMS website.
RMS is engaged by its clients to make referrals to buyers or sellers of private securities (“Securities”). If such client closes a Securities transaction with a buyer or seller so referred, RMS is entitled to a success fee from the client. Such success fee may be in the form of cash or in warrants to purchase securities of the client or client’s affiliate. RMS or RMS representatives may hold equity in its issuer clients or in the issuers of securities purchased or sold by the parties to a transaction.
This communication is confidential and is addressed only to its intended recipient. This communication does not represent an offer or solicitation to buy or sell Securities. Such an offer must be made via definitive legal documentation by the seller of securities.
Investments in the Securities are speculative and involve a high degree of risk. An investor in the Securities should have little to no need for liquidity in the foreseeable future and have sufficient finances to withstand the loss of the entire investment.
RMS does not recommend the purchase or sale of Securities. Potential buyers or sellers of the Securities should seek professional counsel prior to entering into any transaction.
Comments