QumulusAI
NVIDIA GPU cloud and data-center operator for AI workloads.
Something off on this page? Send us feedback.
Profile
Rents NVIDIA GPU compute for AI training and inference through on-demand marketplaces and direct reserved contracts. Pairs a distributed U.S. data-center and power footprint with a legacy cryptocurrency-hosting business that is transitioning toward HPC.
Offers HPC Cloud Services through NVIDIA GPU bare-metal servers and virtualized environments for AI training, fine-tuning, and inference. Capacity is sold on demand through marketplaces and through reserved, multiyear direct contracts, while the legacy cryptocurrency-hosting operation is being allowed to wind down as capacity shifts toward HPC.
Combines Data Center Infrastructure & Hosting with Power & Energy Integration, using leased and colocated capacity, power arrangements, grid-connected sites, and planned behind-the-meter generation. Its distributed U.S. footprint places workload-matched GPUs in smaller sites rather than new hyperscale campuses.
Depends on power, networking, colocation, specialized GPU availability, equipment financing, and marketplace partners; RunPod remains a major demand-generation channel without a purchase minimum.
Company facts
Key statistics
Year to date −62.8% · Average volume (30d) 554.6K
Analysts
No analyst consensus on record.
Earnings
EPS
Revenue
History
| Report date | EPS est | EPS | Surprise | Revenue est | Revenue | Surprise | Next session |
|---|---|---|---|---|---|---|---|
| Aug 25, 2026 | −$0.35 | −$0.72 | −105.7% | $3.2M | $6.7M | +107.5% | −2.71% |
Financial highlights
Not reported.
Dividends
No dividends on record.
Ownership
Shares outstanding 32.87M · Float 15.92M · 48.4% free float
Insider tradesForm 4
Nothing filed in the last 12 months.
View detailsPeers
- CoreWeave
Direct AI-native GPU cloud rival for dedicated and on-demand capacity.
- Amazon
Hyperscale GPU compute alternative for training and inference workloads.
- Microsoft
Azure GPU virtual machines contest enterprise AI compute deployments.
- Alphabet Class C
Google Cloud GPU capacity overlaps in training and inference workloads.
- Oracle
OCI's GPU bare metal and clusters compete for AI infrastructure demand.
- Cerebras Systems
Wafer-scale AI infrastructure offers a non-NVIDIA inference alternative.
- IBM
Enterprise GPU and hybrid AI cloud alternative with multiple accelerators.
CompareAfter-hours close
| Company | Price | Change | |||||
|---|---|---|---|---|---|---|---|
| QumulusAI | $6.17 | +0.98% | $200.8M | — | — | — | — |
| CoreWeave | $84.03 | −0.27% | $45.95B | — | — | 6.1 | +112.3% |
| Amazon | $266.10 | −0.11% | $2.87T | 21.4 | 28.5 | 3.7 | +19.6% |
| Microsoft | $513.06 | −0.06% | $3.81T | 28.6 | 26.2 | 11.5 | +17.8% |
| Alphabet Class C | $342.50 | −0.08% | $4.16T | 17.2 | 25.5 | 9.4 | +24.2% |
| Oracle | $150.95 | +0.17% | $434.14B | 25.9 | 18.7 | 6.4 | +20.6% |
| Cerebras Systems | $178.89 | −0.05% | $40.57B | — | — | 59.6 | — |
| IBM | $235.80 | +0.10% | $221.96B | 20.8 | 18.4 | 3.2 | +1.1% |
| Median | 21.4 | 25.5 | 6.4 | +20.1% |