Skip to content

Croz raises $3.9 billion for mobile computing factories, valuation exceeds $30 billion and Spark factories break construction limits

Share
Croz raises $3.9 billion for mobile computing factories, valuation exceeds $30 billion and Spark factories break construction limits

Listen to this article

Read by Anchor

Data center developer Croz-o closed a Series F financing round of $3.9 billion, lifting its market valuation to $30.9 billion, a move that reflects accelerating institutional capital flows into solid AI infrastructure. The large financing round was led by Atrides Management, Badala Capital, and Valor Equity Partners, with broad participation that included Qatar Investment Authority, Nvidia, Founders Fund, Singapore sovereign wealth fund GIC, as well as Radical Ventures and TPG. This cash infusion arrives just ten months after the company raised $1.38 billion at a $10 billion valuation in October, highlighting the record rise in valuations of specialized power and computing providers.

The company intends to allocate the new proceeds to fund existing data-center projects, including its major facility in Abilene, Texas, dedicated to running OpenAI models, while expanding the construction of modular, small-scale AI factories called “Spark”.Croz-o manufactures the mobile “Spark” units within its own facilities, transports them by truck, and connects them to large power sources wherever they are available.This enables the efficient and rapid deployment of computing capacity without the need to mobilize large construction crews on site. The distributed architecture also overcomes a growing obstacle for conventional data-center developers, namely local community opposition to building massive complexes near population centers.

Croz-o bolstered its governance by announcing three new members to its board: Thomas Saifert, CFO of Cloudflare; Bill Steine, partner and chief investment officer at Primary Digital Infrastructure; and G.P. Strobel, founder and CEO of Redwood Materials and a Tesla board member. Strobel had personally invested in the company in 2021, before Croz-o became the first commercial client for his firm's energy-storage business. The company’s growth rests on a three-track business model that includes leasing data-center space to customers who own their graphics processors, leasing company-owned graphics processors, and selling dedicated compute capacity for running inference models, a model that has attracted customers such as Meta, Microsoft and Oracle.

The round follows Croz-o’s signing of a massive cloud contract worth $13 billion for five years to supply the quantitative trading firm Gen Street with infrastructure and processors, as reported by Bloomberg, while earlier press reports said the company had held meetings with major investment banks such as Goldman Sachs and Morgan Stanley to explore a possible initial public offering. This development marks a shift for the firm, which was founded in 2018 as a cryptocurrency-mining operation using waste natural gas from power fields, before redirecting its focus entirely toward AI as processing demand grew, based on its founders’ view that leading the era of technical abundance requires tight control of the infrastructure from electrons to extracted code.

The shift toward fast modular compute units backed by Gulf capital charts an alternative path for data-center equivalence in the Arab region.The involvement of sovereign investment funds from Abu Dhabi and Doha in financing compute infrastructure demonstrates a practical effort to secure an advanced position in global processing supply chains rather than relying solely on traditional cloud services. For technology leaders and systems-engineering teams in the Gulf and Egypt, the success of the ready-made unit model shows that processing modules can be linked to excess power near generation fields and clean stations to run local inference capacities far faster than lengthy civil-construction schedules. The variety of consumption options, from reserving chips to purchasing inference capacity as a service, provides operational flexibility that helps startups and financial institutions break global queue constraints and lower direct inference costs.

Don't miss the next story

Subscribe for updates