Alpha Compute Moves to Revenue Generation; $21-23 Million Run Rate – Quarterly Update Report

FY26 largely reflects legacy strategic phases; ALP's live compute operations and contracted revenue provide the more relevant basis for assessing its growth trajectory.

ALPHA-01 is operational with 504 B200 GPUs and $16.1 million of annual contracted revenue, while ALPHA-02 targets late-3Q26 deployment with 576 B300 GPUs.

GAMEE adds 120 million users and a scaled application layer, creating new monetization channels and potential internal demand for ALP's compute infrastructure.

Enzo Villani's appointment as President strengthens oversight of strategy, capital allocation, and execution as ALP scales its compute and GAMEE platforms.

Management's $21 million-$23 million annualized revenue run-rate outlook and $200+ million qualified pipeline support visible near-term scale-up and contract-conversion potential.

Modestly valued as ALP trades at 0.65x annualized run-rate sales versus the 7.52x peer average, leaving room for rerating as revenue, margins, and deployment milestones are delivered.

FY26 financials provide limited insight into the business ALP operates today, with the future trajectory now dependent on contracted AI-compute revenue, deployment execution, and pipeline conversion . The fiscal year ended March 31, 2026 (FY26) captured the transition from legacy biotechnology operations through the digital-asset phase and only the earliest costs associated with establishing the company's GPU infrastructure. Nearly all economically relevant developments including the ALPHA-01 customer contract, receipt of the $7.5 million reservation payment, the GAMEE acquisition, and the transition to a $21 million-$23 million annualized revenue run rate occurred after fiscal year-end. Investors should therefore assess ALP primarily through the operating status and economics of its compute clusters, pace of new contract awards, and ability to convert infrastructure deployments into reported revenue and cash flow.

ALPHA-01 has moved ALP from concept-stage infrastructure development to a live, contracted AI compute operator . The company's first enterprise-scale deployment comprises 504 NVIDIA B200 GPUs housed in a hydroelectric-powered Canadian data center under a two-year, $32.2 million agreement with a frontier AI research laboratory. The cluster is fully operational and processing customer workloads, supporting $16.1 million of annual contracted revenue versus approximately $30,000 of quarterly compute revenue before the contract. ALP also received the full $7.5 million upfront reservation payment, strengthening cash-conversion visibility over the 24-month term. The agreement is the first commercial proof point that ALP can source capacity, secure an enterprise customer, deploy a Blackwell cluster, and monetize it under a long-duration infrastructure arrangement.

Multi-year leasing contracts provide recurring revenue visibility while reducing dependence on variable customer utilization . ALP operates as a GPU infrastructure landlord, financing or leasing hardware, securing data-center capacity and power, and then contracting dedicated compute capacity to customers under two- to five-year agreements. Pricing is generally based on an hourly compute rate, but customers commit to reserved capacity and guaranteed uptime, resulting in recurring monthly revenue irrespective of whether the GPUs are used at 100% throughout the period. Recent negotiations have increasingly focused on four- and five-year contracts, which would secure revenue for most of the approximately five-year economic life of the equipment and reduce residual-value exposure. Software and firmware services provide an incremental revenue opportunity where customers require a complete confidential-compute stack rather than bare-metal infrastructure alone.

Targeted contract margins of 20%-25% establish an initial framework for evaluating compute economics as reported revenue begins to scale . ALP is pursuing two primary deployment structures: leasing an existing server cluster and re-leasing the capacity to an enterprise customer at a positive spread, or purchasing servers, financing them through a project-level vehicle, and deploying them at a colocation facility. Across both structures, the company is targeting margins of approximately 20%-25%, with owned-server transactions currently indicating annual returns of roughly 25%-30%. Owned deployments require larger upfront deposits and carry greater exposure to financing rates, power costs, maintenance, colocation expense, and hardware values, but they also provide greater control and higher potential returns. As ALP develops a deployment record and expands its asset base, lower financing costs and improved procurement terms could support incremental margin upside beyond the initial project economics.

ALPHA-02 is the next major deployment milestone and is expected to approximately double ALP's operating Blackwell fleet . The Sweden cluster will comprise 576 NVIDIA B300 GPUs across 72 Dell XE9780 servers and remains targeted for commissioning in 3Q26, taking the combined ALPHA-01 and ALPHA-02 footprint above 1,080 GPUs across Canada and Sweden. The servers have an aggregate cost of approximately $40.9 million, of which $8.6 million had been deposited as of March 31, leaving approximately $32.3 million of remaining hardware consideration. The initial data-center agreement covers 1,213 kW under a 60-month term, with a fixed monthly colocation payment of approximately €163,700 plus variable electricity expense estimated at €0.126 per kWh. ALPHA-02 will therefore validate ALP's ability to coordinate hardware delivery, project financing, colocation readiness, power availability, networking, installation, customer contracting, and commissioning within the targeted quarter.

Asset-backed, non-recourse financing is central to ALP's ability to scale infrastructure without relying primarily on corporate equity . The company generally expects to contribute approximately 20%-30% of deployment cost, with lenders financing roughly 70%-80% against the GPU equipment and associated contracted cash flows. This structure is intended to align debt service with customer payments while limiting recourse to ALP's broader balance sheet. The Vertical Data-related opportunity, subject to closing conditions and lender review, is expected to support approximately $43 million of additional financing for procurement and deployment of B300 capacity. The ability to pair customer deposits, long-duration offtake contracts, and project-level debt will directly influence dilution, return on invested capital, and the speed at which ALP can convert its pipeline into operating clusters.

ALPHA-03 and ALPHA-04 extend the multi-cluster roadmap . The company continues to describe both clusters as in process and expects them to build on the 504-GPU Canadian and 576-GPU Swedish foundation. However, the initial Phase II right of first refusal covering another 1,017 kW at the Sweden facility has expired, meaning future expansion at that location will require a renewed agreement or alternative capacity. ALP is evaluating additional data-center locations, long-term power contracts, real estate, energy rights, and potentially greater ownership across the infrastructure stack. Over time, vertical integration could reduce third-party cost leakage, improve control over deployment schedules and power availability, and support margins above the initial 20%-25% target, although these initiatives will require greater scale and additional project capital.

The $200+ million qualified pipeline reflects late-stage enterprise opportunities with multi-year contract visibility . Potential customers include frontier AI research laboratories, large enterprises, hyperscalers, sovereign entities, and government-related users seeking either dedicated bare-metal capacity or a complete confidential-compute deployment. Individual opportunities generally begin at approximately 500 B200 or B300 GPUs and $16 million-$20 million of annual revenue, while larger inquiries range from 2,000 to 10,000 GPUs. ALP defines qualified opportunities as transactions that have progressed into advanced commercial negotiations involving master service agreements, service-level agreements, final pricing, contract terms, or legal redlines. Management expects a high conversion rate given the maturity of these discussions, although no specific percentage or closing timetable was provided, making signed contracts, customer deposits, and financed deployment plans the clearest measures of pipeline quality.

Additional enterprise contracts, ALPHA-02 commissioning, revenue conversion, and greater reporting transparency are the primary milestones over the next two quarters . A second and third customer agreement would provide stronger evidence of repeatability than the initial $32.2 million ALPHA-01 contract alone, while successful ALPHA-02 deployment would approximately double the operating fleet. Reported results should also begin to demonstrate how the $16.1 million annual contracted revenue converts into recognized revenue, gross profit, operating cash flow, customer receivables, and deferred revenue. ALP currently reports semiannually as a British Virgin Islands foreign private issuer, with the next formal reporting period covering the six months ending September 30, 2026. The company is evaluating more frequent disclosure, potentially including quarterly or selected monthly operating updates, which would improve visibility into deployment progress and financial conversion.

Customer acquisition has initially been driven by technical credibility, the Cocoon relationship, senior industry networks, and organic inbound demand rather than a large conventional enterprise-sales organization . ALP's work on Telegram's Cocoon confidential-inference network provided an early reference architecture and generated inquiries from enterprises and developers seeking similar privacy-focused deployments. The company has since expanded business development toward regulated industries, government customers, sovereign entities, and AI research laboratories where data protection, compliance, and model confidentiality represent material barriers to using shared-cloud infrastructure.

Confidential compute remains ALP's primary competitive differentiator in regulated AI infrastructure . The company combines dedicated GPU capacity with firmware, software, trusted execution environments, and third-party attestation to support end-to-end encrypted AI workloads. The technology layer is being developed through Telegram's Cocoon protocol and ALP's Shroud developer access point, providing customers with the ability to independently verify that sensitive data and models remain isolated from ALP, hardware vendors, and other third parties. This architecture is targeted toward financial services, healthcare, government, defense, and other regulated industries that may be unable to place sensitive workloads in conventional shared-cloud environments. ALP's bare-metal deployment model also provides customers with dedicated infrastructure and greater control over data residency and privacy, although additional enterprise contracts will be required to demonstrate that confidential-compute functionality consistently supports higher pricing, stronger margins, or longer contract terms.