Crusoe Raises $3.9B at $30.9B Valuation

Crusoe, a data center developer, has raised $3.9 billion in a Series F round. Its valuation reached $30.9 billion. According to reporting by Marina Temkin of TechCrunch AI, the financing is positioned as a large-scale deal reflecting expanding demand for AI infrastructure. The company plans to advance large-scale sites and portable small-scale facilities in parallel.

The round was co-led by Atreides Management, Mubadala Capital, and Valor Equity Partners. Existing and new investors including Founders Fund, GIC, Nvidia, Qatar Investment Authority, Radical Ventures, and TPG participated. Three new directors are also joining the board. They are Thomas Seifert, Chief Financial Officer of Cloudflare, Bill Stein, Partner and Chief Information Officer at Primary Digital Infrastructure, and JB Straubel, Founder and Chief Executive Officer of Redwood Materials.

This article is based on TechCrunch AI (All Rights Reserved) and relies on fair quotation under Article 32 of the Japanese Copyright Act.

Background to the $3.9B Raise and Confirmed

Valuation

Crusoe announced the completion of its $3.9 billion raise by September 18. Its valuation was calculated at $30.9 billion. About 10 months earlier, in October of the previous year, it had raised $1.38 billion at a $10 billion valuation. Its valuation has thus roughly tripled in a short period.

Notably, sovereign funds and a semiconductor giant joined alongside the three co-leads. Participation by public funds such as GIC and Qatar Investment Authority stands out. Nvidia’s participation is seen as illustrating the link between supply chains and compute demand. This further strengthens its positioning as an AI infrastructure company.

The board appointments also signal a focus on strengthening its foundation. Seifert brings expertise in operating and financing large-scale cloud infrastructure. Stein is a specialist in telecom infrastructure investment. Straubel, who also serves as a Tesla director, is an expert in energy storage technology.

Straubel’s relationship with Crusoe goes back some time. He invested in Crusoe personally in 2021. Crusoe later became the first customer for Redwood’s energy storage business. The appointments reflect an approach that treats power and compute infrastructure as one.

How the Mobile AI Factory Spark Works and Its Aims

At the core of the use of proceeds is advancing existing plans. A representative example is the large-scale site in Abilene, Texas. The site is reported to be used by OpenAI. It is infrastructure supporting both large-scale training and inference demand.

The other pillar is portable small-scale AI factories. They have been announced under the name Spark. They are manufactured in the company’s own factory and designed to be transported by freight truck. They can operate anywhere once connected to a large-scale power source.

In-house manufacturing shortens the time to deployment. There is no need to assemble a large construction workforce on site. Greater flexibility in securing power is another advantage. Units can be deployed in a distributed manner near demand centers or power sources.

This approach advances the standardization of the construction process. Enclosures, cooling, power, and networking are treated as a single integrated product. It reduces variability in on-site work and stabilizes quality. The design also allows for common maintenance parts.

Methods for validating computing environments are also growing in importance in terms of standardizing operating infrastructure. For example, there are testing infrastructure initiatives such as GNOME OS Test Center, Inspired by Apple TestFlight. In image-generation infrastructure, SenseTime’s 8B Image Generation Model SenseNova U1.5 Lite Officially Released has been published. In operational monitoring, COSMIC Epoch 1.1 Featuring New System Monitor ‘COSMIC-Monitor’ offers a useful reference. They share the idea of managing distributed computing resources.

Three-Tier Revenue Structure and Analysis of

Large Contract Trends

Crusoe’s revenue structure consists of three pillars. The first is leasing facility space. It houses customer-owned GPUs and supplies them with power. The second is leasing its own GPUs. The third is selling compute capacity needed to run models. It captures demand for processing known as inference.

The three-tier structure helps stabilize utilization rates. Facility leasing lends itself to long-term contracts. Equipment leasing commands high unit prices but is exposed to demand fluctuations. Inference sales generate usage-based revenue. The combination increases resilience to economic cycles.

The existence of large contracts is also drawing attention. According to Bloomberg, the company has signed a five-year, $13 billion cloud computing contract for quantitative trading firm Jane Street. It covers the supply of GPUs and AI infrastructure. The scale is exceptional for a single customer.

Its customer base includes Meta, Microsoft, and Oracle. It has touchpoints with both hyperscalers and enterprise customers. Its equity relationship with Nvidia could contribute to stability in procurement. There is room for collaboration on both securing equipment and sales.

The co-founder and chief executive officer outlined the company’s direction in a statement.

controlling the infrastructure from electrons to tokens, and we’re grateful to have investors who share that conviction. The idea is to consistently control infrastructure from electricity to compute tokens. He expressed gratitude for sharing that conviction with investors.

Strategy for Local Opposition and Power

Constraints

Developing large-scale facilities is creating friction in many areas. In the United States, a survey found that more than 60% of voters oppose construction nearby. Concerns over noise, landscape, and power consumption are behind the opposition. Cases of prolonged local permitting are also increasing.

Small-scale distributed units like Spark could help ease such friction. They can add capacity without building huge new structures. They can be installed adjacent to existing power facilities. They are designed to limit physical impact on local communities.

Power constraints are a more structural challenge. Power consumption continues to grow as training and inference workloads expand. Expanding the transmission grid takes a long time. Siting close to power plants offers a competitive advantage.

The relationship with Redwood can be interpreted as groundwork on the power front. Combining energy storage can absorb load fluctuations. It also contributes to continued operation in emergencies. Power procurement and operation can be optimized as one.

IPO Speculation and the Journey Since Founding

Axios has reported moves toward a public listing. Crusoe has met with investment banks including Goldman Sachs and Morgan Stanley. They reportedly discussed a public offering in the near future. It has not been announced as an official decision.

About 10 months have passed since the previous raise. A large follow-on raise in a short period indicates strong demand. It also shows the scale of capital needs for facility investment. A listing would diversify fundraising options.

The company was founded in 2018 as a cryptocurrency mining business. It used surplus gas from flaring of associated gas at oil fields as a power source. It later pivoted to AI infrastructure amid surging demand for compute. It repurposed its expertise in securing power for compute infrastructure.

Growth since the pivot has been rapid. Its customer base and contract scale continue to expand. This capital increase will be used to expand production capacity and supply systems. The direction toward in-housing manufacturing and operations is clear.

Editorial Opinion

Looking at the short-term impact. We expect this capital increase to accelerate competition in facility supply over the next three to six months. With mass production of Spark in addition to Abilene-scale large sites, backlog fulfillment will speed up. As a company holding leverage over equipment procurement and power contracts, we assess that Crusoe’s negotiating power will further increase. Competitors will be forced to respond on delivery times and siting flexibility. Competition over pricing and uptime guarantees could intensify.

From a long-term perspective, we expect distributed infrastructure to become a standard option in one to three years. The shift will move from reliance on giant facilities alone to a combination of portable units and large sites. As grid constraints persist, deployment close to power sources is a rational solution. Combining with energy storage will contribute to operational stability. We also expect approaches to building consensus with local communities to change.

We pose questions from the editorial team. Whether distributed deployment will truly ease local opposition remains to be seen. Even at smaller scale, total power consumption and noise may remain unchanged. Whether reliance on a specific customer at the $13 billion level is healthy is also at issue. Vertical integration controlling everything from electricity to compute tokens is efficient. On the other hand, is there not also concern about increasing concentration on specific providers?

References

• “Crusoe raises $3.9B to build massive data centers and small modular “AI factories””, by Marina Temkin — TechCrunch AI, 2026-09-17T23:25:52.000Z (ARR)
• Source URL: https://techcrunch.com/2026/09/17/crusoe-raises-3-9b-to-build-massive-data-centers-and-small-modular-ai-factories/