Digi Power X closes billion-dollar deal with Cerebras Systems for 40 MW AI campus in Alabama
Digi Power X and Cerebras Systems just closed one of the most talked-about deals in the AI infrastructure sector in 2026. The announcement, made on May 5, put the Nasdaq-listed company under the ticker DGXX at the center of financial market and tech ecosystem attention almost instantly.
The company signed a 10-year Master Services Agreement with Cerebras, initially valued at $1.1 billion and with the potential to reach $2.5 billion when renewal clauses are factored in. On top of that, the contract includes an expansion option that could add another $1.4 billion to the total deal value.
The core of the contract is the construction and operation of an AI data center campus with 40 MW of capacity in Columbiana, Alabama. The project was designed from scratch to support the high-density computational workloads required by cutting-edge AI acceleration hardware, following Tier III infrastructure standards.
The market reaction was explosive. 📈 DGXX shares surged more than 62% right after the announcement, peaking at 69.2% within less than an hour of trading. Trading volume was 8.7 times above average, and the move added roughly $106 million to the company’s market cap, which climbed to around $275 million.
This is the kind of event that transforms a low-profile company into a front-page story, and understanding what’s behind this colocation agreement helps explain why so many people had their eyes on this one. 👀
What this colocation agreement is and why it matters so much
A colocation agreement is basically a contract where one company provides physical space, electrical power, cooling, and network infrastructure so another company can install and operate its own computing equipment. In the case of Digi Power X and Cerebras Systems, the arrangement goes well beyond simple rack rental — this is about building a dedicated campus, engineered from the ground up to meet the specific demands of high-density AI workloads.
Every detail of the project, from power distribution to the cooling system, was planned to handle the kind of intense processing that Cerebras chips require, without the bottlenecks typical of general-purpose data centers. That’s critical when we’re talking about hardware like the Wafer Scale Engine, which demands far greater thermal density than conventional servers.
What makes this colocation agreement especially noteworthy is the scale and the commitment baked into the arrangement. A $1.1 billion contract with a 10-year term and renewal clauses that could push the total value to $2.5 billion doesn’t come along every day in this sector. It signals a long-term commitment between the two companies, with Digi Power X stepping into the role of strategic infrastructure partner for Cerebras as it expands in the AI computing market.
This kind of arrangement reduces operational uncertainty for both sides: Cerebras secures predictable, dedicated processing capacity, while DGXX locks in recurring revenue for an entire decade. For investors, that predictability is one of the most valuable assets an infrastructure company can offer.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Construction phases and delivery timeline
The Columbiana campus will be developed in two clearly defined phases, with an aggressive timeline that reflects the AI market’s urgency for additional computational capacity:
- Phase 1: 15 MW of IT load, expected to be operational by December 15, 2026
- Phase 2: An additional 25 MW of IT load, with completion scheduled for the end of Q1 2027
Digi Power X stated that Phase 1 construction will begin immediately, reflecting the company’s readiness in terms of power, site development, and long-lead equipment procurement. Phase 1 is being funded with the company’s own resources, which shows confidence in the return on investment and the long-term value of the Cerebras partnership.
Revenue generation is expected to begin by the end of 2026, with the full billing ramp kicking in after the complete delivery of all 40 MW in Q1 2027. This timeline positions the company to start converting contractual value into actual cash flow in less than two years from the announcement date.
Why Columbiana, Alabama, was the chosen location
Columbiana might not be the first place that comes to mind when you think of tech hubs in the United States, but the choice makes a lot of sense from a large-scale data center infrastructure perspective. The region offers a combination of factors that are becoming increasingly decisive for projects of this size:
- Access to robust and reliable power infrastructure
- A favorable regulatory environment for data center construction and operation
- Proximity to major fiber optic corridors serving the southeastern United States
- Availability of land for large-scale construction at competitive costs
One of Digi Power X’s most significant advantages in this project is that the company owns the land where the campus will be built. This provides a balance sheet-backed development platform, setting the company apart from competitors that rely on leasing arrangements to operate.
On top of that, the company has already completed construction of a dedicated substation on-site to serve Phase 1, with grid interconnection finalized and a power supply agreement in place with Alabama Power. These two factors — guaranteed power and an on-site substation — eliminate two of the biggest development risks typically associated with large-scale data center projects. It’s the kind of preparation that allows for a significantly accelerated construction timeline.
Who is Cerebras Systems and why they needed this deal
Cerebras Systems is one of the most ambitious companies in the AI hardware ecosystem. Founded in 2016 and headquartered in Sunnyvale, California, it became known for developing the Wafer Scale Engine, an AI chip that takes up an entire silicon wafer — something unprecedented in the industry. This design enables far greater processing density than conventional GPUs for certain types of models, especially large language models and large-scale neural networks.
The company has been growing rapidly and counts research labs, governments, and major tech companies among its clients — organizations looking for alternatives to the Nvidia-dominated AI acceleration market.
To sustain that growth, Cerebras needs robust and reliable physical infrastructure, and that’s exactly where the Digi Power X deal comes in as a strategic piece of the puzzle. Building and operating proprietary data centers requires capital, operational expertise, and time — resources that a company in rapid expansion mode would rather direct toward research, development, and sales than facilities management. By signing a long-term colocation agreement with DGXX, Cerebras secures scalable processing capacity without having to become an infrastructure operator itself.
It’s also worth noting that Cerebras is at a sensitive moment in terms of public visibility. The company filed for its IPO with the SEC back in 2024, but the process was paused due to regulatory issues. Closing a billion-dollar contract with a Nasdaq-listed partner reinforces the narrative of operational strength and once again draws the attention of investors who were waiting for signs that the company remains on a consistent growth trajectory.
What Digi Power X represents in this new cycle of AI infrastructure
Digi Power X is a relatively new player in the AI data center space, but this contract completely changes the tier at which the market perceives the company. Before the announcement, the company had a low-volume profile on Nasdaq, with pretty quiet trading activity. The stock explosion on the day of the announcement shows that the market interpreted this deal as a real turning point for the company.
The company describes itself as a vertically integrated AI infrastructure operator, with a portfolio of energy assets and data center capacity spread across Alabama, New York, and North Carolina, totaling approximately 400 MW of secured power across its sites. Additionally, the company’s NeoCloudz platform offers GPU-as-a-Service on dedicated Nvidia bare metal infrastructure.
The 40 MW campus project in Columbiana already puts Digi Power X in a different league within the AI infrastructure segment. High-density data centers built for AI workloads are scarce assets that are expensive to build. The combination of dedicated power, precision cooling, and low-latency connectivity requires specialized engineering and heavy investment from the planning stage onward.
The fact that the company managed to structure a project with these characteristics and still attract a client of Cerebras Systems’ caliber for a decade-long colocation agreement indicates that there’s real technical and commercial competence behind this operation.
The company’s recent track record and path to the billion-dollar deal
This Cerebras contract didn’t come out of nowhere. Over the past year, Digi Power X built its AI infrastructure narrative step by step, through a series of milestones that paved the way for a deal of this magnitude:
- November 2025: The company strengthened its liquidity to roughly $85 million and unveiled operational capacity targets of up to 195 MW
- Early 2026: Updates on the ARMS 200 system commissioning, planning for 10 MW in Alabama with projected revenue of $15 million per MW, and the launch of the US Data Centers platform valued at approximately $10 million
- April 20, 2026: Signing of the company’s first AI revenue contract, valued at $19.6 million over 24 months, for GPU leasing
- May 5, 2026: Announcement of the Master Services Agreement with Cerebras, initially valued at $1.1 billion
This progression shows a company that scaled its commitments in a structured way — from balance sheet strengthening and infrastructure buildout to increasingly larger revenue contracts. The leap from a $19.6 million contract to a $1.1 billion one is dramatic, but it didn’t happen disconnected from an operational foundation that was already being put together.
The company also has the financial flexibility to support its growth. In April 2026, an S-3 shelf registration approved by the SEC allows Digi Power X to issue up to $750 million in various types of securities, including an at-the-market program of up to $75 million in subordinate voting shares. This mechanism provides flexibility to fund the Tier III data center construction and associated power infrastructure as needed.
What company executives said about the deal
Michel Amar, Chairman and CEO of Digi Power X, described the deal as transformational for the company, highlighting that signing a $1.1 billion anchor contract with a top-tier AI computing company validates everything the company has built — including its team, its sites, its infrastructure capabilities, and its vision for what a next-generation data center operator should be.
Alec Amar, President of Digi Power X, emphasized that closing a deal of this magnitude with one of the most prominent emerging companies of the AI era positions Digi Power X as a serious player operating at the highest level, and that this type of transaction can open doors to new sophisticated tenants, lenders, and partners.
Hans Vestberg, Senior Advisor at Digi Power X and former Chairman and CEO of Verizon Communications, compared the challenge of building high-density AI infrastructure to one of the defining projects of the current generation — on the same scale as the 4G and 5G network rollouts that transformed global connectivity.
The broader landscape of AI infrastructure demand
Looking at the sector as a whole, this deal is yet another sign that demand for infrastructure dedicated to artificial intelligence is far from slowing down. Companies of all sizes are racing to secure computational capacity, and the specialized colocation model is gaining traction as a viable alternative between building data centers from scratch and relying exclusively on public cloud providers.
Training and running inference on large language models consumes enormous amounts of energy and requires specialized hardware with thermal requirements that traditional data centers simply weren’t designed to handle. This is creating an entirely new class of infrastructure, with construction standards, cooling systems, and power distribution architectures that look completely different from what existed just five years ago.
Digi Power X positioned itself well within this window of opportunity, and the Cerebras contract is the most concrete proof of that. If the Columbiana project stays on schedule — with Phase 1 operational in December 2026 and all 40 MW delivered by the end of Q1 2027 — the company could solidify its presence as one of the emerging go-to names in AI data centers across the United States. 🚀
For anyone following the AI infrastructure market, this is a story worth keeping an eye on over the coming months — especially the construction milestones and any announcements of new contracts with other clients that could be drawn to the Columbiana campus.
