What happened with QMLS today?
No read yet
Uses news and trading volume to explain the day’s move and how it relates to the market.
Stop guessing. Get the call, a three-month target price, the best and worst cases, and a source for every claim.
QumulusAI supplies computing resources to machine-learning teams, AI startups, research institutions and enterprises that need GPU capacity. It installs and maintains servers in leased data centers, with most compute sales coming through marketplaces rather than direct customer contracts. RunPod supplies most of its GPU-service revenue: RunPod contracts with users, sets prices and collects payments, then passes an agreed share to QumulusAI. Most usage is on short-term or on-demand terms. RunPod has no minimum purchase obligation and can end the relationship on notice.
Compute services supplied most revenue in the latest reported quarter, but bitcoin mining and mining hosting remain substantial businesses. Hosting clients pay QumulusAI to operate and maintain their mining equipment, including charges covering electricity with a markup. Hosting revenue comes almost entirely from a few customers, including Fortitude Mining. QumulusAI also supplies computing power to Luxor Technology Corporation’s mining pool and receives bitcoin payments. It plans to move capacity into AI computing as hosting contracts expire without renewal.
The business depends on available electricity, leased data-center capacity and access to GPU equipment from a limited supplier base. Expansion requires substantial spending and additional financing; its USD.AI financing arrangement does not guarantee future funding. U.S. export controls and sanctions restrict where and with whom it can do business and may constrain equipment sourcing.
The risk section is the longest, dullest part of the filing. This keeps only what can really hurt you: where it hits, how hard, when it plays out.
Every company claims an edge. Each one is checked against hard numbers like margins and returns: which hold, and which line gives way first.
QumulusAI’s GPU-cloud results hinge on privately held RunPod, which generated 71.9% of its GPUaaS revenue in the first quarter of 2026 and controls customer acquisition, pricing and billing for that capacity.
CoreWeave competes with QumulusAI by selling GPU cloud capacity and software that runs and manages AI workloads, but targets larger enterprise and AI-lab deployments with longer commitments. Its large-scale customer contracts offer a contrast to QumulusAI’s mostly short-term, usage-based business.
Corvex competes with QumulusAI for customers paying to train and run AI models without operating their own computing infrastructure. Both offer managed GPU systems and direct access to dedicated servers, so Corvex’s AI cloud business—not its legacy healthcare operations—is the relevant comparison.
SharonAI Holdings, Inc. competes with QumulusAI by selling on-demand GPU capacity to customers training AI models, running inference and doing research. Its cloud platform is based in Australia, making its GPU-as-a-Service business a geographically distinct comparison.
QumulusAI depends on NVIDIA’s GPU technology to sell AI cloud compute, so GPU availability and new hardware generations matter to its expansion and economics. Watch B200 and B300 availability: those chips were expected to make up about 80% of QumulusAI’s planned GPU mix for the end of 2026.
Applied Digital and QumulusAI both depend on demand for powered AI data centers as they shift resources from blockchain hosting to high-performance computing. Applied Digital leases and hosts facilities for large cloud providers and other compute customers, rather than selling GPU cloud compute; its progress securing power and grid connections is the useful signal for QumulusAI’s expansion.
Oracle competes with QumulusAI for organizations paying for cloud computing and high-performance computing to run AI workloads. Oracle offers a broader general-purpose cloud, while QumulusAI focuses on customers needing smaller, customized GPU deployments.
Super Micro Computer and QumulusAI are both exposed to the pace of AI capacity investment: Super Micro sells AI and high-performance computing servers to cloud providers, while QumulusAI installs GPU infrastructure and charges customers to use it. Super Micro’s server business serving cloud providers is the relevant comparison, rather than its broader enterprise and edge offerings.
Microsoft’s Azure competes with QumulusAI for customers buying cloud and high-performance computing, with a broader service range and scale that give those customers more alternatives. Microsoft says large data centers lower computing costs per unit, making Azure’s pricing and bundled services worth watching against QumulusAI’s specialized GPU offering.
Amazon’s AWS competes with QumulusAI for customers paying for cloud infrastructure and high-performance computing, and its breadth and scale put pressure on QumulusAI’s pricing and differentiation. AWS—not Amazon’s retail operations—is the business to compare with QumulusAI’s customized GPU cloud offering.
Alphabet’s Google Cloud competes with QumulusAI to sell cloud infrastructure and high-performance computing to organizations, but its advertising business makes the stocks a less direct comparison. Watch Google Cloud separately from Google Services, which earns mainly from advertisers on Search, YouTube and other properties.
CoreWeave, Inc. earningsCRWV
CoreWeave’s report will show whether larger customers are committing to its GPU cloud, a useful contrast for QumulusAI, which earns most of its compute revenue from flexible usage.
Read CoreWeave’s GPU-cloud growth alongside its customer commitments. If both are strong, customers are locking in capacity for longer, supporting demand for QumulusAI’s GPU cloud but not proving its mostly short-term users will make the same commitments. If growth is weak or commitments limited, GPU demand may be harder to turn into durable revenue.
Nvidia Corp earningsNVDA
NVIDIA’s report will offer clues about GPU supply and new-chip rollouts, which QumulusAI needs to expand its AI computing capacity.
Watch NVIDIA’s supply and rollout of new GPU generations. More supply and smooth deliveries would suggest fewer obstacles to QumulusAI obtaining chips for HPC expansion, but would not show that QumulusAI received an allocation or favorable pricing. Tight supply or delays would raise the risk of slower deployments or higher costs.
Applied Digital Corporation earningsAPLD
Applied Digital’s report will show progress securing power and grid connections for AI data centers, a useful infrastructure signal for QumulusAI’s HPC expansion.
Watch Applied Digital’s progress securing power and grid connections for its data centers. Smooth progress would suggest the infrastructure needed to expand AI computing is becoming easier to obtain, a favorable signal for QumulusAI’s shift toward HPC, but it would not confirm power for QumulusAI’s own sites. Delays would point to power and grid access still holding back new capacity.
