Runpod compute deals
Public agreements, delivery targets, and source evidence for Runpod.
Market overview
The commitments behind the buildout
Selected public agreements. Amounts follow the selected filters. Compute purchases and the facilities hosting them can describe the same capacity.
| Layer | Stated amount | Up to | More than |
|---|---|---|---|
| Compute purchases | — | — | — |
| IT power leases | — | — | — |
| Campus projects | — | — | — |
Upper limits are not firm commitments. Minimums are not exact totals. Each column stands alone; the three layers must not be added together.
USD compute agreements
—
Stated contract values over their full terms
Additional upper-limit values
—
Additional minimum values
—
0 agreements with comparable USD values. Excludes financing, construction budgets, monthly fees, and overlapping contract updates. Full-term contract values are not annual revenue or infrastructure capital expenditure.
Compare buyers · 1 companies
| Buyer / developer | Agreements | Compute: stated | Compute: up to | Compute: over | Dated targets |
|---|---|---|---|---|---|
| Runpod | 1 | — | — | — | 1 of 1 |
Time to Power
When is capacity expected?
1 dated targets · 0 without a first-capacity window
0 elapsed targets await delivery evidence
Delivery confirmation is still needed for these records. An elapsed target alone does not establish a delay.
Actual time to power: not established
Company targets describe expected availability. Measured time to power requires a dated start and site-level energization evidence.
Reported delivery
QumulusAI / Runpod · MW undisclosed · all 616 gpus active; deployment completed in august ahead of the september 1 target
QumulusAI reports all 77 nodes active across two one- and two-year reserved-capacity agreements. It does not disclose MW or where within its US data-center footprint the GPUs are hosted.
2026
Delivery reported
Site not disclosed · All nodes active in August 2026; QumulusAI reports completion ahead of the September 1 target
First compute capacity · A first phase may cover only part of the agreement.
How to read these dates and figures
Dates are the original announced windows. A vague year keeps a full-year range; construction completion and equipment release dates are excluded. 1 announcements disclose no MW allocation. 0 quantified records are excluded from capacity comparisons because their scope is not comparable or overlaps. Export the CSV for the source, quantity terms, and inclusion basis of each agreement.
1 announcement
| Announced | Buyer / partner | Value | Capacity | Location / timing | Details |
|---|---|---|---|---|---|
| Sep 3, 2026 | Runpod Runpod Cloud compute | Undisclosed | Undisclosed 616 NVIDIA RTX PRO 6000 Blackwell GPUs across two reserved-capacity agreements; MW and customer contract value undisclosed | Location undisclosed All nodes active in August 2026; QumulusAI reports completion ahead of the September 1 target Delivery reported | In use at announcementDeal detailsQumulusAI reports full deployment of 616 GPUs for Runpod in August 2026, ahead of the September 1 target. The agreements have different one- and two-year terms, but value and MW are not disclosed. QumulusAI says the GPUs are served from its existing US data-center footprint; no specific site or utility energization date is given. Hardware: 616 NVIDIA RTX PRO 6000 Blackwell GPUs in 77 nodes QumulusAI reports Runpod deployment completion ↗Assess this deal |