Digi Power X’s $1.1 Billion AI Contract Dwarfs Its $6.6 Million Q2 Revenue

Digi Power X reported $6.6 million in second-quarter revenue while promoting a $1.1 billion AI infrastructure contract. That contrast pushed the company into google news, but it also defines the central question surrounding its AI pivot.

The contract is real, material, and unusually large compared with Digi Power X’s present operations. However, most of its associated revenue depends on infrastructure that is still under construction in Columbiana, Alabama.

Digi Power X must deliver the first 15 megawatts of contracted capacity by December 2026. It then plans to expand the Cerebras campus to 40 megawatts during the first quarter of 2027. Until those milestones arrive, investors are comparing current revenue with a much larger promise.

This is not simply another quarter from a small data center operator. It is an early test of whether a former cryptocurrency infrastructure business can become a dependable AI landlord.

The opportunity is clear. AI companies need power, cooling, network capacity, and facilities that can support dense computing systems. Digi Power X already controls energy infrastructure that can shorten part of the development process.

The risk is equally clear. A power-connected property is not automatically a commissioned AI data center. Digi Power X still needs to complete construction, install specialized equipment, satisfy service conditions, and bring customer capacity online.

Google News Focuses on the Contract, but Q2 Revenue Shows the Starting Point

Digi Power X’s reported Q2 revenue remains small beside the value assigned to its Cerebras agreement.

The Quiver Quantitative item circulating through google news highlighted $6.6 million in quarterly revenue and the $1.1 billion AI contract. Those figures describe different periods of the company’s development.

Quarterly revenue measures services that Digi Power X delivered during the three months ending June 30. The contract value represents expected payments across an initial ten-year term, subject to the company meeting its obligations.

Dividing the headline contract value across ten years suggests a much larger future revenue base than the current quarter supports. That comparison is useful for scale, but it is not a forecast for evenly distributed revenue.

The underlying agreement uses a phased development plan. It calls for 15 megawatts of IT load in Phase 1, followed by another 25 megawatts in Phase 2.

The complete contract filing identifies Cerebras Systems as the customer. It describes a purpose-built facility capable of supplying 40 megawatts of IT load with Tier III infrastructure.

Tier III refers to a data center design with redundant capacity components and maintainable systems. The designation matters because AI accelerators require consistent power and cooling during maintenance or equipment failures.

Digi Power X has targeted December 15, 2026, for the first phase’s ready-for-service date. It expects the full 40-megawatt deployment by the end of the first fiscal quarter of 2027.

Those dates establish a straightforward test. The company must turn a construction project into working capacity before the contract can reshape reported revenue.

The Q2 figure therefore should not be treated as evidence that the full Cerebras economics have already arrived. The quarter ended before either contracted phase was scheduled to enter service.

That timing also explains why the headline appears more dramatic than the income statement. Digi Power X signed the agreement in May, but signing did not immediately create ten years of accounting revenue.

Revenue recognition normally follows the delivery of contracted services. For this project, that requires operational data center capacity, not merely a signed document.

The agreement nevertheless changes the company’s position. A named customer can support financing conversations, equipment commitments, and construction planning more effectively than speculative demand projections.

It also introduces customer concentration. One relationship will carry substantial weight if the campus becomes the dominant source of AI colocation revenue.

That concentration is not automatically negative. Large data center projects often begin with anchor tenants that justify the initial build.

However, it means execution problems affecting one facility or customer can have an outsized effect. Digi Power X’s small existing revenue base offers limited insulation while the campus is being completed.

The proper reading of Q2 is therefore neither purely optimistic nor dismissive. The reported revenue shows how early the transition remains, while the contract establishes what successful execution would change.

The Cerebras Agreement Turns Construction Into the Main Contest

The primary conflict is no longer cryptocurrency mining versus AI demand. It is Digi Power X’s contractual promise versus the physical work required to deliver it.

Digi Power X entered the colocation agreement with Cerebras on May 4, 2026. Colocation means the operator supplies secured space, power, cooling, and related infrastructure for equipment owned or controlled by a customer.

The arrangement differs from selling short bursts of GPU computing through an online cloud platform. It also differs from cryptocurrency mining, where revenue depends heavily on token prices, network difficulty, and energy costs.

A long-term colocation agreement can produce more predictable revenue once the facility is operating. The customer commits to capacity, while the operator accepts development and service obligations.

Digi Power X describes the contract’s initial value as approximately $1.1 billion. It says the potential value reaches approximately $2.5 billion if Cerebras exercises a seven-year extension.

The larger figure is not the current contract term. It depends on an extension that has not occurred and on Digi Power X continuing to satisfy its obligations.

The company’s first-quarter filing explains the two-stage timetable. Phase 1 covers 15 megawatts, while Phase 2 brings the total to 40 megawatts.

That filing also shows how the agreement fits inside a broader transition. Digi Power X historically earned revenue through cryptocurrency mining, energy sales, and conventional colocation.

Its new strategy adds several AI infrastructure layers. These include purpose-built colocation, modular data center systems, and GPU access through the NeoCloudz platform.

NeoCloudz is the company’s GPU-as-a-Service operation, which rents access to graphics processors for AI workloads. Digi Power X says the platform began generating AI-related revenue in May 2026.

The first operating deployment uses NVIDIA Blackwell GPUs inside the company’s ARMS 200 modular unit. ARMS stands for AI-Ready Modular Solution, a prefabricated system intended to accelerate data center deployment.

This smaller installation matters because it gives the company an operating reference before the entire Cerebras campus is complete. It can test cooling, power distribution, networking, and service procedures at a limited scale.

However, the modular unit should not be confused with the 40-megawatt contract. A live GPU cluster validates some operating capabilities, but it does not prove that the larger campus will arrive on schedule.

The Cerebras project requires two distinct achievements. Digi Power X must first complete a much larger facility, then operate it according to the agreement’s service requirements.

Construction risk includes equipment availability, utility coordination, contractor performance, commissioning, and unexpected site conditions. AI facilities also require specialized electrical and cooling systems because accelerator racks can draw far more power than conventional servers.

Commissioning is especially important. This process tests whether the facility’s systems function together under normal loads and failure scenarios.

A building shell can look nearly finished while critical testing remains incomplete. Investors should therefore focus on accepted capacity rather than photographs of construction progress.

The company reported in July that crews were erecting the building shell. It also said commitments were in place for major long-lead equipment needed for Phase 1.

Those are useful progress indicators. They reduce some procurement uncertainty, but they do not replace a completed ready-for-service test.

A Former Mining Operator Is Selling Its Access to Power

Digi Power X’s advantage is not a new AI model or accelerator. It is control over sites, power connections, and operating infrastructure.

AI data centers face a basic physical constraint. Chips can be manufactured faster than utilities can approve new high-capacity electrical connections in many markets.

Digi Power X entered the AI infrastructure race with properties already tied to energy production and industrial computing. Its portfolio includes sites in Alabama and New York, plus a longer-term development project in North Carolina.

That background explains the company’s pivot. Cryptocurrency mining and AI computing use different processors, but both require large electrical loads, cooling, networking, and around-the-clock facility operations.

The overlap is meaningful, though incomplete. Bitcoin miners usually operate specialized ASIC machines, while AI systems rely on GPUs or other accelerators connected through high-speed networks.

AI customers also expect stricter uptime, security, latency, and redundancy standards. Converting a mining site therefore requires more than replacing one type of computer with another.

Digi Power X’s strategy combines existing energy access with new Tier III infrastructure. The company is betting that its power position can reduce the time needed to secure operational AI capacity.

This approach follows a wider industry pattern. Cryptocurrency operators have increasingly marketed powered properties to AI companies seeking locations for high-density computing.

Core Scientific’s hosting relationship with CoreWeave became an important reference for that model. Other operators, including IREN, Hut 8, TeraWulf, and Cipher Mining, have also pursued AI or high-performance computing opportunities.

The sector shift began because miners possessed something AI developers urgently needed: large power allocations attached to operational industrial sites.

Digi Power X is smaller than several of those peers. That makes the Cerebras agreement more consequential relative to its present business.

It also makes financing and construction discipline more important. Larger operators can spread risk across multiple campuses, customers, and operating contracts.

Digi Power X has tried to create several paths into AI revenue. The Cerebras campus provides long-term colocation, while NeoCloudz offers direct GPU computing services.

A separate agreement with SubQ AI covers dedicated access to NVIDIA Blackwell systems through NeoCloudz. Digi Power X described that arrangement as a 24-month contract that became effective in May 2026.

These businesses expose the company to different risks. GPU rental revenue depends on equipment utilization, customer demand, and the pace of hardware obsolescence.

Colocation transfers more hardware risk to the customer but leaves the operator responsible for the facility. The operator must maintain power, cooling, availability, and other contracted services.

The Cerebras project is therefore the cleaner test of Digi Power X’s infrastructure thesis. It does not require the company to win a broad public cloud market before producing meaningful revenue.

Instead, it asks whether the company can serve one major AI computing customer at a purpose-built campus. Success would support the claim that its energy assets can become higher-value digital infrastructure.

Failure would show the limit of the mining-to-AI comparison. Existing power access helps, but it does not eliminate the engineering, financing, and operational work required for an enterprise data center.

What the $1.1 Billion Headline Does Not Guarantee

The contract supplies revenue visibility only after Digi Power X delivers the contracted capacity and keeps it available.

Large contract values often receive more attention than the conditions attached to them. That effect is amplified when a deal dwarfs the supplier’s recent annual revenue.

The $1.1 billion figure covers an initial ten-year term. It should not be read as cash already received, revenue already recognized, or a guaranteed near-term return.

Digi Power X’s obligations include constructing, equipping, and commissioning both phases. The company must also meet the operational requirements contained in its agreement with Cerebras.

The agreement contains redactions for confidential commercial terms. Public filings therefore give investors the overall structure without revealing every service metric, remedy, or pricing detail.

That information gap matters. Data center contracts can include acceptance tests, performance standards, service credits, termination provisions, and conditions tied to phased delivery.

Investors can verify the total contracted power and target dates. They cannot independently model every circumstance that might delay or reduce revenue from the public document alone.

Digi Power X’s July operations update projected that the colocation agreement would contribute approximately $80 million to $100 million during 2027.

The company said that estimate reflects a full year of Phase 1 operations and a partial year of Phase 2. It remains company guidance, not reported revenue.

The same update said Phase 1 remained on track for December 2026. Phase 2 was expected by the end of the first fiscal quarter of 2027.

These projections create measurable checkpoints. They also leave little room for readers to treat construction delays as secondary details.

A delay in Phase 1 would affect the first stream of material contract revenue. A Phase 2 delay would reduce the amount of capacity contributing during 2027.

Capital requirements deserve similar attention. AI data centers need electrical gear, cooling equipment, networking, building work, and commissioning services before customer payments can scale.

Digi Power X reported a much stronger liquidity position during 2026 than it had maintained historically. Even so, investors should watch how much cash construction consumes and whether new financing dilutes existing shareholders.

The company has maintained an at-the-market equity program. Such programs let a business issue shares gradually through public markets.

That flexibility can help fund development. It can also increase the share count, which spreads future earnings across more owners.

Customer concentration creates another risk. Cerebras is not simply one customer among many within the company’s current revenue base.

The contract can become Digi Power X’s defining asset if both phases enter service. Any renegotiation, dispute, customer strategy change, or acceptance delay would therefore carry substantial consequences.

Cerebras itself operates in a competitive AI accelerator market. It sells wafer-scale computing systems and cloud capacity against an industry shaped heavily by NVIDIA hardware and large cloud providers.

Its need for additional data center capacity supports Digi Power X’s opportunity. However, the customer’s expansion strategy also connects the project to changing demand for AI compute.

The Q2 headline should therefore be read as a statement about backlog and planned capacity. It is not proof that the company has completed its transition.

Google news exposure can bring a small company wider attention. It cannot shorten an equipment lead time or complete a commissioning test.

Three Signals Will Show Whether the AI Pivot Is Working

The next two reported quarters should replace much of the current narrative with observable operating evidence.

The first signal is Phase 1’s ready-for-service status. Digi Power X has targeted December 15, 2026, for delivery of the first 15 megawatts.

A completed and customer-accepted Phase 1 would strengthen the central case behind the contract. It would move the Cerebras relationship from construction backlog into operating infrastructure.

Investors should look for precise language. “Construction complete,” “energized,” and “ready for testing” do not always mean that a customer has accepted service.

The strongest disclosure would confirm that contracted capacity entered service and began generating colocation revenue. It should also state whether the milestone occurred on schedule.

A delay would not automatically invalidate the entire project. Large data center developments regularly encounter schedule changes.

However, a delay without a revised commissioning plan would weaken management’s 2027 revenue expectations. Repeated timetable changes would place even more weight on financing and customer remedies.

The second signal is Phase 2 construction and equipment readiness. That phase adds 25 megawatts, bringing the Cerebras campus to its full 40-megawatt contracted load.

Digi Power X has targeted the end of the first fiscal quarter of 2027 for full deployment. Investors should watch procurement, utility work, building completion, and commissioning progress.

The company has said major long-lead equipment for Phase 1 is secured. Similar clarity about Phase 2 would reduce uncertainty around the full build.

Phase 2 determines whether the campus becomes a large operating asset or remains a smaller first-stage installation. It also affects how much of management’s projected 2027 revenue range is achievable.

The third signal is the changing composition of reported revenue. The company’s $6.6 million Q2 figure does not yet demonstrate a large AI infrastructure business.

Future filings should separate legacy activities from NeoCloudz and Cerebras-related revenue wherever reporting rules permit. That breakdown will show whether AI revenue is adding to the business or simply replacing declining operations elsewhere.

The mix also matters for margins. Energy sales, cryptocurrency mining, GPU rentals, and long-term colocation have different cost structures.

Revenue growth without improving gross economics would tell a less convincing story. Investors should compare revenue with power costs, depreciation, operating expenses, and construction-related cash use.

NeoCloudz provides an earlier indicator while the Cerebras campus is being built. The company says its ARMS 200 unit has generated AI revenue since May 15, 2026.

The size and durability of that contribution remain important. A small initial deployment can establish technical credibility, but it must attract continuing workloads to become a meaningful business line.

Readers should also watch Cerebras. Its own cloud growth and infrastructure spending can influence how urgently it needs the Alabama capacity.

Recent AI cloud results showed strong growth in Cerebras cloud revenue alongside weaker hardware performance. That mix can increase the strategic importance of dependable computing capacity.

Still, customer growth does not remove supplier risk. Digi Power X must finish the facility regardless of how compelling broader AI demand appears.

The most useful question is no longer whether AI data centers attract capital. It is whether Digi Power X can convert contracted demand into accepted capacity without losing control of its schedule or balance sheet.

The Contract Is the Beginning, Not the Q2 Result

Digi Power X has secured a credible route to a much larger business, but its current numbers still describe a company in transition.

The $1.1 billion Cerebras agreement gives Digi Power X something many speculative AI infrastructure projects lack: a named customer and a filed contract.

It also gives readers clear milestones. Phase 1 is supposed to deliver 15 megawatts in December 2026, followed by 25 additional megawatts during the first quarter of 2027.

Those commitments make the story easier to evaluate. The company either completes the capacity, passes the required tests, and begins recognizing material revenue, or it does not.

The reported $6.6 million Q2 revenue figure supplies a useful baseline. It prevents the contract headline from being mistaken for the company’s present operating scale.

At the same time, judging the contract solely through that quarter would ignore its development schedule. The quarter ended months before the first contracted phase was due.

Digi Power X has already taken several steps beyond a paper pivot. Its modular Alabama installation is live, NeoCloudz has begun generating revenue, and the Cerebras agreement identifies a substantial customer requirement.

Yet the most valuable capacity remains unfinished. Construction progress, customer acceptance, and cash management now matter more than repeated descriptions of AI demand.

That is the reversal behind the google news headline. The large number does not settle the investment case; it makes execution easier to measure.

Developers and enterprise AI buyers should care because the project illustrates where the compute bottleneck is moving. Access to accelerators matters, but access to commissioned power and cooling increasingly determines when those accelerators can operate.

Infrastructure operators should care because Digi Power X is testing whether former mining assets can meet enterprise AI standards. Success would support similar conversions at power-rich sites across North America.

Investors should care because the contract changes the company’s potential scale without removing its development risk. Backlog becomes valuable only when the operator delivers service under acceptable economics.

The next meaningful update should therefore answer specific questions. Did Phase 1 remain on schedule, did the customer accept the capacity, and did AI revenue become visible in the financial statements?

Until then, the most defensible conclusion is balanced. Digi Power X has a significant agreement and a plausible infrastructure strategy, but neither has completed the transition reflected in its market narrative.

Readers following the company through google news should look past the contract total and track the three operating signals. Watch customer acceptance, full campus commissioning, and the share of revenue produced by AI services.