Lambda

Lambda, founded in 2012 in San Francisco by brothers Stephen and Michael Balaban, builds GPU cloud services and on-premise AI hardware for training and inference, positioning itself as a 'superintelligence cloud.' The company sells on-demand and reserved GPU clusters, private cloud, and physical GPU workstations and servers, and is among NVIDIA's backed cloud partners. Lambda raised over $1.5 billion in a Series E led by TWG Global in November 2025, following a $480 million round in early 2025 at a roughly $2.5 billion valuation that included NVIDIA. The company is reportedly targeting an IPO in the second half of 2026, and its roughly $350 million pre-IPO convertible-note raise is now reported closed, led by Mubadala Capital with conversion at about a 20% discount to the IPO price. Revenue for the fiscal year ended September 2025 was reported above $520 million, heavily concentrated in Microsoft and Nvidia.

One of the longest-running independent GPU clouds, Lambda spans both cloud and on-prem hardware and is a frequently-cited IPO candidate in the AI-infrastructure wave.

lambda.ai AI / GPU Cloud

United Statesprivate

At a glance

Total raised

$2.4B

last round series e

Headcount

201-500

approximate · public range

Headquarters

San Francisco

United States

Founded

2012

14 years active

Status

Private

AI / GPU Cloud

Patents on file

For subscribers

Subscribe to see the full profile

Including Funding rounds, Bull / Bear thesis, Stock + earnings, Roster changes, Patents, News, and Open roles.

Already subscribed? Sign in →

AI bull / bear

Bull case

Lambda is the neocloud with a fourteen-year operating history: founded in 2012 by Stephen and Michael Balaban, it was selling GPU infrastructure before the transformer era and pairs a cloud business with an on-prem hardware line — workstations and servers — that gives it a customer-acquisition channel no pure rental cloud has. The numbers now match the ambition. Lambda has raised roughly $2.36B, including a $1.5B+ Series E led by TWG Global and USIT in November 2025 on top of an early-2025 $480M round at about $2.5B that included NVIDIA, and reported revenue passed $520M for the fiscal year ending September 2025 with third-quarter sales up 80% year over year. Anchor demand is contractual, not speculative: a multibillion-dollar November 2025 agreement with Microsoft covering tens of thousands of GPUs including GB300 NVL72 systems, and a four-year, ~$1.5B arrangement in which NVIDIA leases back roughly 18,000 of its own GPUs. The path to liquidity is concrete — Morgan Stanley, JPMorgan, and Citi are mandated, a ~$350M pre-IPO convertible led by Mubadala Capital is in place, and a 2026 listing would make Lambda the second pure-play AI cloud on public markets, with CoreWeave's multiple as the comp and a cleaner balance-sheet story.

Bear case

Strip out two customers and Lambda's growth story thins dramatically. Microsoft reportedly accounted for around 62% of 2024 revenue, and its largest single customer today is NVIDIA — leasing back its own GPUs in a circular arrangement that flatters revenue while adding no independent demand signal; an S-1 will put that concentration in print at the worst possible moment. The pre-IPO financing terms read as urgency rather than strength: the ~$350M convertible is priced at a reported ~20% discount to the eventual IPO and carries financial penalties if Lambda fails to list within a year — a structure that transfers timing risk onto the company just as commentators describe the IPO environment as unfriendly. Competitively, Lambda is mid-sized in a market that punishes the middle: it has raised $2.4B against Nebius's $44B+ contracted backlog, Nscale's $14.6B valuation and 200,000-GPU Microsoft pipeline, and CoreWeave's public currency. GPU rental is a depreciation treadmill — every dollar of revenue requires continuous capex on silicon that obsoletes in two to three years — and the hardware business is low-margin ballast. If the IPO window closes or prices below the convertible's assumptions, the penalty mechanics compound into a downward spiral of exactly the kind late-cycle infrastructure companies rarely escape.

Generated by editorial-sourced · July 1, 2026Refreshes quarterly · see disclosures