The Balaban brothers' AI cloud, valued by Forbes at about $2.5 billion in 2025, has closed a $1 billion credit facility and priced a separate $926 million loan as it considers a public listing.
By RuntimeWire Staff · Published
Primary source: Bloomberg Technology
Why it matters #
Bloomberg's reported terms would value Lambda at almost five times Forbes's September 2025 estimate, setting a demanding benchmark for the Balaban brothers as they consider taking the capital-intensive AI cloud provider public.
Lambda is weighing up to a $3 billion financing at a valuation of $12 billion or more, Bloomberg reported August 24. Brothers Stephen Balaban and Michael Balaban founded the San Francisco AI cloud infrastructure provider in 2012.
The proposed terms remain under negotiation. Bloomberg did not identify a lead investor, and Lambda has announced neither an IPO filing nor a formal timetable. The financing could position Lambda for a possible 2027 public listing and would extend a capital buildout that has moved Lambda far from its beginnings as two machine-learning engineers trying to reduce their cloud-computing bill.
Stephen now serves as chief technology officer, while Michael is chief product officer, according to Lambda's leadership page. Their original insight came from experience: machine-learning projects consumed enough cloud capacity to undermine the economics of the products they were building.
From Noisebridge to AI infrastructure
Stephen studied computer science and economics at the University of Michigan and became the first engineering hire at Perceptio, which developed neural networks that ran on iPhone GPUs before Apple acquired it. Michael studied computer science and discrete mathematics at Michigan, then worked on Nextdoor's infrastructure team.
The brothers started Lambda after their machine-learning projects made AWS compute costs prohibitively expensive. They concluded that $60,000 of owned equipment could replace much of that recurring expense, and Michael said the move cut their AWS bill by 95%. The brothers encountered the same problem in their own machine-learning work: prohibitively expensive AWS bills pushed them to build a GPU cluster and eventually sell access to other researchers, according to The Next Platform.
That experience became Lambda's commercial thesis. Other researchers faced the same mismatch between rising GPU requirements and general-purpose cloud pricing, so the brothers began selling machines and compute access. Lambda now offers individual GPU instances, clusters and dedicated infrastructure for AI training and inference.
By September 2025, Lambda had approximately 15,000 paying customers and an estimated valuation of about $2.5 billion, according to Forbes. Those figures make the proposed $12 billion-plus valuation almost five times the earlier estimate, although the timing and terms of the two valuations differ.
Lambda's company overview says it serves AI developers, enterprises, hyperscalers and frontier-model labs, with infrastructure that can scale from one GPU to hundreds of thousands. The business has moved from serving individual researchers and developers toward larger GPU deployments for enterprise and AI-lab workloads.
The founders brought in infrastructure operators
Lambda's preparations for that larger job became explicit on May 5, 2026, when Stephen Balaban moved from CEO to CTO and Lambda appointed Michel Combes as chief executive. Combes previously ran Sprint, Alcatel-Lucent and SoftBank Group International. Lambda also named former AT&T Communications CEO John Donovan as chairman and moved Michael Balaban into the chief product officer role.
In Lambda's announcement of the appointments, Lambda also identified Jerry Hunter, Charles Fisher and David Connolly as recent additions.
Stephen retained responsibility for technology strategy and architecture, preserving the founder's role in the product while handing daily operations to an executive whose career was built in capital-intensive telecommunications. In the same announcement, Lambda said the expanded team was working toward 3 gigawatts of AI compute under management by 2030. That figure is a Lambda target rather than independently verified operating capacity.
The leadership change gives Combes the tasks that tend to precede an infrastructure IPO: financing construction, expanding financial controls and persuading investors that GPU fleets can generate durable returns after depreciation, replacement and borrowing costs.
Lambda is financing each layer separately
The proposed round follows Lambda's $320 million Series C in February 2024. Lambda announced its $480 million Series D in 2024, followed by more than $1.5 billion from TWG Global and USIT in November 2025.
US Innovative Technology Fund led the Series C, with B Capital, SK Telecom, T. Rowe Price, Crescent Cove, Mercato Partners, 1517 Fund, Bloomberg Beta and Gradient Ventures participating. Andra Capital and SGW co-led the Series D, joined by Nvidia, ARK Invest, In-Q-Tel, Andrej Karpathy, Fincadia Advisors, G Squared and existing investors. Lambda's November 2025 release said TWG Global led that financing, with USIT and existing investors participating.
Debt has become another major component. On May 7, 2026, Lambda announced that it had closed a $1 billion syndicated senior secured credit facility. The multi-tranche facility expanded a $275 million line established in August 2025, with J.P. Morgan acting as lead arranger. Lambda said the financing could fund Nvidia accelerator infrastructure and data-center capacity.
On August 12, 2026, Lambda priced a separate $926 million senior secured first-lien term loan B facility. Lambda's release said the collateral included GPU servers and related infrastructure, with contracted cash flows from compute services supporting the loan.
Lambda's primary release said the term loan was priced at the secured overnight financing rate plus 3 percentage points and issued at 99.5%. The loan carries a Baa2 rating from Moody's, matures on December 31, 2030, and is scheduled to amortize fully. Lambda identified Morgan Stanley as lead-left arranger and MUFG as joint bookrunner. The release said the loan was expected to close in August 2026, subject to customary closing conditions; the supplied reporting does not establish that it had closed as of August 25.
Lambda has disclosed at least one large contract behind its expansion. On November 3, 2025, Lambda announced a multibillion-dollar, multi-year agreement with Microsoft covering deployments with tens of thousands of Nvidia GPUs.
CoreWeave already set the comparison
Lambda operates in the category commonly called neoclouds: providers built around specialized AI computing instead of a broad catalog of general-purpose cloud services.
CoreWeave emphasizes large, multi-year contracted deployments. Crusoe and Nebius also compete in AI cloud infrastructure, while AWS, Microsoft Azure and Google Cloud remain the incumbents with established customer relationships and balance sheets.
The proposed valuation depends on Lambda showing that its customer base and large deployment contracts can support the capital required for expansion. McKinsey's analysis of neoclouds identifies depreciation, electricity costs, customer concentration and refinancing as category-wide pressures. Those are sector risks, not reported Lambda financial results.
A completed $3 billion round would give Lambda more purchasing power before facing public-market scrutiny. At a valuation of $12 billion or more, investors would be betting that the founders can convert contracted demand and a reported 15,000-customer base into returns after lenders, power providers and hardware suppliers have been paid.