Lambda Is Raising $3B at a $12B+ Valuation — the Neocloud Wave Heads for the Public Markets

Lambda, the Nvidia-backed AI cloud provider, is in talks to raise as much as $3 billion at a valuation of $12 billion or more, Bloomberg reported Monday. The round is explicitly structured as pre-IPO financing, positioning the company for a public listing next year.

The trajectory is steep. Lambda raised more than $1.5 billion in November in a round led by TWG Global. Nine months later it is negotiating a valuation roughly double that round’s, with projected 2026 revenue north of $1.5 billion. Its backer list reads like an AI-industry cap table bingo card: Nvidia itself, Andra Capital, Scott Hassan’s family office SGW, OpenAI co-founder Andrej Karpathy, and Cathie Wood’s ARK Invest.

Lambda belongs to the “neocloud” cohort — providers that rent raw GPU capacity and AI infrastructure rather than the full managed-services buffet of AWS or Azure. CoreWeave proved the model can IPO; Nebius rode the same wave. Lambda’s differentiation has always been developer-first positioning: it started selling deep-learning workstations to researchers in 2012, long before “GPU cloud” was a category, and its on-demand pricing remains one of the most legible in the market.

The debt behind the equity

The equity raise is only half the financing picture. Two weeks ago Lambda sold a $917 million loan to fund an Nvidia-tied chip purchase — the now-standard neocloud playbook of borrowing against GPU collateral and long-term customer contracts. That structure works beautifully while H200s and Vera Rubin racks hold their value. It gets uncomfortable if chip generations turn over faster than the loans amortize, which is precisely the scenario Nvidia’s accelerating release cadence makes plausible.

That is the tension buyers of the eventual IPO will have to price. Neoclouds are capital-intensive middlemen between Nvidia and AI labs, and Nvidia sits on both sides of the table — as supplier, as investor, and increasingly as the financier of its own demand. Nvidia backing $105 billion for OpenAI’s Ohio data center and taking equity in its resellers is a flywheel, until it is a circular dependency.

Why developers should care

Practically: more funded neoclouds means continued downward pressure on GPU-hour pricing, and Lambda has historically been aggressive there. If you rent A100s, H100s, or B200s for training or fine-tuning, competition between Lambda, CoreWeave, Nebius, and the hyperscalers is the reason spot pricing keeps drifting down even as demand grows.

The IPO signal matters too. A successful Lambda listing next year would confirm public markets are still open to pure-play AI infrastructure at scale — and give the entire private AI stack, from labs to chipmakers, a fresh valuation benchmark. Terms are not final, and the deal could still change shape. But the direction is unmistakable: the picks-and-shovels layer of AI is going public, one neocloud at a time.

Sources

• Bloomberg
• Yahoo Finance