Cerebras Systems continues to strengthen its ability to deploy large-scale artificial intelligence infrastructure. The company has signed a ten-year colocation agreement with CleanCore Solutions to occupy an entirely new data center campus in Minnesota, a facility that will feature 55 MW of electrical power and 40 MW of critical load for AI systems once completed. According to the announced terms, the contract has an initial value of approximately $800 million, with two renewal options that could raise the potential total above $3 billion.
The key details of Cerebras’ new campus in 20 seconds
- Cerebras will lease 100% of an AI campus in Minnesota for an initial period of ten years.
- The center will have 55 MW of electrical capacity and 40 MW of critical IT load upon completion.
- The contract has an initial value of around $800 million and could exceed $3 billion with the planned renewals.
- The project is part of Cerebras’ strategy to expand its global inference network and computing capacity for AI.
The announcement comes just days after Cerebras confirmed its intention to deploy 200 MW of data center capacity in Europe before the end of 2027, with projects in Norway, Finland, and France. The company is building an international network of proprietary and partnered infrastructure to deliver AI inference services with lower latency and higher availability across different markets.
A campus designed for AI from the ground up
The Minnesota complex has been designed according to Tier III standards, a common architecture for critical facilities that allows maintenance tasks without service interruptions.
When completed, the campus will feature:
| Feature | Value |
|---|---|
| Total electrical capacity | 55 MW |
| Critical IT load | 40 MW |
| Electrified capacity to date | 20 MW |
| Expected occupancy | 100% for Cerebras |
| Estimated start of operations | First quarter of 2027 |
| Initial contract duration | 10 years |
According to information provided by CleanCore Solutions, around 20 MW already has power supply, enabling an initial phase with approximately 15 MW of IT capacity, while the rest of the infrastructure is completed over the next few months.
Cerebras’ expansion accelerates its global strategy
While NVIDIA continues to dominate the AI accelerator market, Cerebras is pursuing a different strategy.
In addition to developing its Wafer Scale Engine (WSE) processors, the company is building an international network of data centers optimized for inference and model training.
In recent weeks, the company has announced:
- The expansion of its infrastructure in Europe.
- New deployments in the United States.
- An increase in capacity for enterprise clients and foundational model developers.
Its goal is to provide computing capacity where customers are located, reducing dependence on centralized infrastructure.
The AI boom continues to reshape the colocation market
The agreement also reflects a sector trend: increasingly, developers prefer to secure capacity through long-term contracts even before data centers are operational.
With electricity availability becoming a major industry bottleneck, reserving power for ten or twenty years is becoming a common practice among AI providers, hyperscalers, and cloud operators.
The Minnesota campus will be fully occupied from its launch, providing revenue visibility for the developer and ensuring Cerebras prioritized access to energy resources in a market where megawatt availability is becoming more limited.
A deal noteworthy also for its protagonist
Beyond the numbers, there’s another noteworthy aspect: CleanCore Solutions was not until recently an operator of digital infrastructures.
In 2026, the company announced a strategic shift to gradually move away from its traditional cleaning products business and focus on developing infrastructure for artificial intelligence. It also announced a partnership with HST Technologies to develop data center campuses and launched a financing program to support this transition.
This move illustrates how the rise of AI is attracting new players to the data center market, a sector where energy availability, land, and financing have become strategic assets.
The race is no longer just about making chips
For years, AI competition focused mainly on designing more powerful accelerators.
Now, the focus also includes owning the infrastructure needed to power them.
Companies like Cerebras, NVIDIA, OpenAI, Microsoft, Meta, and xAI compete to secure electrical capacity, land, cooling systems, and long-term contracts to deploy new hardware generations.
In this context, agreements like the one in Minnesota show that infrastructure is becoming as critical as the processors themselves.
Frequently Asked Questions
What capacity will Cerebras’ new campus have?
The project features 55 MW of electrical power and 40 MW of critical IT load upon completion.
What is the value of the contract?
The initial deal is approximately $800 million over ten years, with renewal options that could surpass $3 billion.
When will it start operating?
CleanCore Solutions estimates that revenue generation and operations will begin during the first quarter of 2027, as the campus is progressively commissioned.
How does this project fit into Cerebras’ strategy?
It is part of their international AI infrastructure expansion, which also includes the recent announcement of 200 MW capacity in Europe before the end of 2027.
via: prnewswire
