QumulusAI is a cloud infrastructure company specializing in rapid deployment of graphics processing unit (“GPU”)-powered solutions for artificial intelligence (“AI”) applications, serving a critical market that is often overlooked by large-scale cloud providers (“hyperscalers”), which operate massive, standardized computing infrastructures primarily serving the largest enterprises. Our platform delivers flexible, competitively priced, and customizable solutions for underserved small and mid-market customers—including machine learning teams, AI infrastructure startups, and research institutions—while also supporting the scale and complexity requirements of large enterprises, such as long-term deployments or supplemental on-demand compute capacity. The Company traces its origins to WAHA Technologies, Inc. (“WAHA”) and WAHA, Inc. (renamed SPRE Commercial Group, Inc., or “SPRE”), both incorporated in 2019. SPRE focused on data center assets and operations, while WAHA specialized in blockchain managed services. In December 2022, the two entities completed a corporate roll-up to form Global Digital Holdings, Inc. and remain wholly owned subsidiaries of the Company. In April 2025 (after a substantial minority investment in October 2023), Global Digital Holdings, Inc. acquired The Cloud Minders, Inc. (a company focused on GPU-as-a-Service (“GPUaaS”) assets and operations), now a wholly owned subsidiary, and rebranded the combined operations as QumulusAI. On August 18, 2025, Global Digital Holdings, Inc. changed its name to QumulusAI, Inc. Today, we distinguish ourselves by deploying, activating and producing revenue from GPU infrastructure within approximately 90 days. We do so across a distributed network of sales channel and platform partners and the strategic use of stranded colocation facilities, enabling us to connect with thousands of developer clients with an expedited path from procurement to revenue generation. Traditional infrastructure providers, by comparison, can take up to 12 to 24 months to activate new capacity. QumulusAI is headquartered in Atlanta, Georgia. We operate in multiple colocation data centers in Marietta, Georgia, Kansas City, Missouri, Denver, Colorado, and Philadelphia, Pennsylvania, with combined deployed and contracted capacity of approximately 7 MW of power. We have secured rights of first refusal for 15 megawatts (“MW”) of information technology (“IT”) load capacity space for our GPU equipment and we are actively planning for expansion exceeding 120 MW of total IT load across our platform with potential to support over 90,000 NVIDIA B200/B300 GPUs (among the latest generation GPUs purpose-built for foundation model training), or as many as 1,500,000 GPUs optimized for AI inference at scale. Additionally, in the aggregate, we operate approximately 10 MW of grid power with immediate access to more than 40 MW of additional grid power in Watonga, Oklahoma and Denton, Texas where we manage blockchain assets. These sites serve as foundational assets for power-intensive compute deployments and provide strategic flexibility for future infrastructure repurposing. At present, the majority of the Company’s active power capacity is allocated to blockchain asset management rather than high-performance computing (“HPC”). As of the date of this filing, the Company operates approximately 10 MW of blockchain load in Watonga, Oklahoma. By comparison, the Company’s HPC business currently utilizes approximately 2.6 MW of IT load (approximately 3.6 MW total power load), supporting approximately 2,136 GPUs currently deployed and an additional 952 GPUs in the process of being delivered and deployed. Although blockchain currently represents the majority of deployed power, the Company’s growth strategy is focused on HPC. In 2026, the Company expects, for the first time, revenue generated from HPC compute services to exceed revenue generated from blockchain operations, reflecting the higher revenue density of HPC workloads relative to blockchain. By the end of 2026, the Company expects its HPC operations to support approximately 11.0 MW of total HPC IT load (approximately 15.4 MW total power load), inclusive of approximately 1,100 GPUs deployed in 2025 and the planned deployment of over 5,800 additional GPUs, consisting of an estimated mix of approximately 80% B200 and B300 data center GPUs and approximately 20% RTX Pro 6000–class servers. With respect to blockchain operations, the Company expects to continue operating approximately 10 MW of blockchain load in Watonga, Oklahoma on an ongoing basis. SPRE TULSA OK, LLC, an indirect wholly owned subsidiary of the Company, sold its interest in the T20 Mining Group, LLC (“T20”) joint venture on February 13, 2026, which supported approximately 50 MW of blockchain load in Tulsa, Oklahoma. This transaction significantly reduced our blockchain operation while providing a $16.5 million cash injection to our balance sheet. The Company intends to use this capital to accelerate the development and scale of its HPC business. Following this transaction, the Company expects that approximately 10 MW of blockchain load in Denton, Texas, will remain scheduled, resulting in approximately 20 MW of total blockchain operations inclusive of the Watonga facility. Separately, the Company has approximately 9 MW of on-grid power availability in Watonga, which it plans to allocate toward HPC data center expansion, representing approximately 6 MW of additional HPC IT load (approximately 8.4 MW total power load). As a result of these planned deployments and portfolio changes, the Company expects that by the end of 2026, a greater portion of its active power capacity will be allocated to HPC compute workloads than to blockchain asset management. While the Company currently expects to have approximately 20 MW allocated or otherwise activated for blockchain operations and approximately 23.8 MW allocated or otherwise activated for HPC compute workloads by that time, actual allocations may vary based on deployment timing, customer demand, power availability, and execution. In addition, the Company is actively evaluating opportunities to secure additional powered land for future HPC data center construction and to expand its data center colocation footprint, although there can be no assurance as to the timing or scale of any such expansion. The address of our principal executive office is in Atlanta, Georgia.
Read More
