From Electrons to Tokens: Crusoe Closes $3.9B Series F at a $30.9B Valuation

The largest private financing in the AI infrastructure sector this quarter just closed. Crusoe, the Denver-based company that brands itself “the AI factory company,” announced the initial closing of its anticipated Series F on Thursday: $3.9 billion raised at a $30.9 billion post-money valuation, in an oversubscribed round co-led by Atreides Management, Mubadala Capital, and Valor Equity Partners.

The headline number is the tripling of valuation in under a year — Crusoe’s October 2025 Series E raised $1.375 billion at just above $10 billion — but the more consequential figure sits further down the announcement: over $140 billion in total contracted value (TCV) across its vertically-integrated platform, and more than 6 gigawatts of gross contracted capacity across data centers and cloud, with over 1 GW already delivered and operational.

What the money buys

Unlike conventional data center developers, who treat power as a constraint to navigate after site selection, Crusoe runs what it calls an energy-first strategy: it originates and manages power directly at the source, pairing in-house power plant development with partnerships spanning grid, battery, nuclear, thermal and renewable energy providers. The new capital is earmarked for scaling that entire stack — from large-scale vertically-integrated campuses down to Crusoe Spark, the modular data centers the company manufactures in the US and treats as its critical growth engine.

Spark is the most strategically interesting line item in the release. Because the units are factory-built, Crusoe says they compress the timeline of data center construction in the field from years to weeks, deliver speed and certainty of capacity energization, and lower costs relative to conventional construction. They are designed for the next generation of silicon and networking systems, and their modularity lets customers grow capacity incrementally as workloads demand. In a market where the binding constraint on AI growth is increasingly the speed at which megawatts can be energized, a factory line for data centers is a direct attack on the bottleneck.

The rest of the round fuels Crusoe Cloud, where bookings are growing at a claimed triple-digit rate year to date, and Managed Inference, which the company says has contracted over a meaningful share of capacity since launching late last year — powered by an inference engine claiming up to 9.9x faster time-to-first-token and 5x higher throughput versus vLLM, built on Crusoe’s MemoryAlloy architecture. External validation is cited too: a Gartner Magic Quadrant Visionary position, #1 for inference speed on Artificial Analysis, and status as an NVIDIA Exemplar Cloud.

The cap table as a signal

The investor list reads like a cross-section of everyone with a structural stake in the compute buildout. Beyond the three co-leads, the round includes Founders Fund, GIC, NVIDIA, the Qatar Investment Authority, Radical Ventures, and TPG — plus a long tail of new names: Altimeter, ARK Invest, Baillie Gifford, Fidelity, T. Rowe Price, Tiger Global, StepStone, Salesforce Ventures, Polychain, Ribbit Capital, and SemiAnalysis Capital, among others.

NVIDIA’s participation is the least surprising name on the list. Crusoe is one of the most prominent “neoclouds” — the class of GPU-aligned infrastructure operators (CoreWeave, Lambda, Nebius, Nscale) that buy Nvidia silicon at scale and rent it to frontier labs and enterprises. Every dollar flowing into neoclouds ultimately flows back to the chipmaker, so NVIDIA investing in its own best customers is now a familiar flywheel.

Gavin Baker, Atreides’ Managing Partner and CIO, framed the thesis in one line: “As AI grows, the economics flow to the lowest-cost producer of intelligence. Through a vertically-integrated model, Crusoe owns the entire value chain — a structural advantage that compounds as they build.”

The OpenAI imprimatur

The release also confirms the customer credential that matters most: OpenAI trained Astra, its first artificial general intelligence, at the Abilene, Texas campus Crusoe designed and built. That 1.2 GW campus — jointly developed with Blue Owl Capital and Primary Digital Infrastructure, with a total project cost around $12 billion — has become the reference site for hyperscale AI construction. Current cloud customers named in the announcement include Cognition, Figure, and Perplexity, the latter following the multi-year full-lifecycle partnership (training on GB300 NVL72 clusters plus Managed Inference) announced just two days earlier.

The timing of the round’s close, one day after the Perplexity deal, is not subtle. Crusoe is stacking evidence — a frontier-lab flagship site, a fast-growing inference business, a modular manufacturing line, and now $3.9 billion in fresh equity — into a single narrative: that owning electrons, buildings, and the cloud layer above them is the only way to keep pace with demand that hyperscalers cannot absorb alone.

Why it matters

The Series F crystallizes a shift in how the AI buildout gets financed. The first wave of neocloud growth was debt-driven: operators borrowed against long-term GPU contracts, as CoreWeave’s convertible notes and Lambda’s investment-grade debut demonstrated. Crusoe’s round shows the equity market is now willing to underwrite the fully integrated version of the thesis — power generation included — at a $30.9 billion price.

It also sharpens the sector’s stratification. Crusoe differentiates from pure-play GPU renters by owning generation assets and from hyperscalers by serving AI-native and frontier customers without competing with them as model providers. Its workforce has crossed 1,800 across five countries, with new offices in Bellevue and New York.

The bet, in CEO Chase Lochmiller’s words: “We believe AI will usher in an era of abundance… Getting there means controlling the infrastructure from electrons to tokens.”

Whether that abundance materializes on schedule is the industry’s open question. What is no longer in question is who is paying for the attempt — and at what price.