Runpod vs Vultr vs Lambda
Compared on what each company has disclosed: named sites, capacity, capital, GPU prices and listing, with a source beside every figure in the profiles.
Side by side
| Measure | Runpod | Vultr | Lambda |
|---|---|---|---|
| Index category | GPU cloud | NeoCloud | NeoCloud |
| Listing | Private | Private | Private |
| Named sites (Capital Atlas) | none on record | 1 | 2 |
| Disclosed capacity, all sites | not stated | 50 MW | 24 MW |
| Operating or building | not stated | not stated | not stated |
| Disclosed site capital | not stated | $1.0B | not stated |
| States with named sites | none on record | OH | MO, OK |
| GPU prices tracked | 7 | 7 | 7 |
| Accelerators priced | 4 | 5 | 3 |
| Lowest H100 list price, per GPU-hour | $2.89 | $2.30 | $3.29 |
| Lowest B200 list price, per GPU-hour | $6.79 | $3.50 | $6.69 |
| Sources cited in the Index | 0 | 2 | 2 |
Runpod
Full profile: Runpod →Vultr
Vultr announced a financing at a $3.5 billion valuation in December 2024 with LuminArx Capital Management and AMD Ventures, with the round size reported as $333 million. In December 2025 it said it would invest more than $1 billion in a 50MW AI cluster in Springfield, Ohio using 24,000 AMD MI355 chips. Anchor customer contract values are n.a.
Full profile: Vultr →Lambda (Lambda Labs)
Lambda has layered a $320 million Series C, a $500 million GPU-backed SPV facility led by Macquarie, a $480 million Series D, a Series E of more than $1.5 billion, and debt facilities including a $926 million term loan B, a $1 billion syndicated facility and a $275 million facility, with a $350 million pre-IPO round reportedly in talks ahead of a targeted H2 2026 IPO. Its anchor relationships include a multibillion-dollar Microsoft agreement covering GB300 NVL72 systems and a reported $1.5 billion Nvidia lease-back. Flagship US expansion is a Kansas City, Missouri deployment starting at 24MW and scaling above 100MW against a company-wide 3GW liquid-cooled ambition. On 31 August 2026 Bloomberg and The Wall Street Journal reported that Lambda had signed a $35 billion cloud agreement with Anthropic, with the capacity to run through Hut 8's Beacon Point campus in Nueces County, Texas on a lease held by Nvidia. The reports did not disclose the agreement's term, GPU count, payment schedule or computing capacity.
Full profile: Lambda →Which is larger, Runpod or Vultr or Lambda?
By capacity disclosed at named sites, Vultr, with 50 MW across 1 sites. Runpod has no megawatt figure on record at 0 sites. Lambda has 24 MW at 2 sites.
Which publishes the lowest H100 price?
Vultr, at $2.30 per GPU-hour on its published list price. Runpod lists $2.89. Lambda lists $3.29.
Where do these figures come from?
Sites, capacity and capital come from the SAVRN Capital Atlas, which keeps a source link beside every figure. Prices come from the SAVRN price tracker, which reads each provider's own page or feed. The company profiles linked above carry every source.
Seven references, one system
Capital, delay, policy, water, grid, permits and scarcity are one system seen from seven sides. Each reference is primary-sourced, refreshed on a schedule, and links into the others, so a project can be followed from the money committed to the docket that stopped it.