Layer 2.5 · head to head
Vultr vs Civo
Geographic spread against Kubernetes simplicity.
| Vultr | Civo | |
|---|---|---|
| Can you leave | Movable with effort. Expect to redo images, storage wiring and networking. | Portable. The workload is standard enough to move to another vendor without a rewrite. |
| Counterparty | Lower than the pure-play neoclouds — the GPU business sits on top of a profitable general hosting business rather than being the whole company. | Small provider; the GPU line is one product among several. |
| What it is | Purpose-built GPU cloud. Owns or leases its own datacentre capacity and sells it directly. | Purpose-built GPU cloud. Owns or leases its own datacentre capacity and sells it directly. |
| How you reach it | You get virtual machines. The familiar cloud model. | You get a cluster. Assumes you already run Kubernetes at scale. |
| Accelerators | H100, H200, A100, L40S, MI300X, plus fractional GPUs | A100, L40S, and NVIDIA datacentre parts |
| Regions | Global, 30+ locations | UK, US, EU |
| Pricing model | Hourly on-demand, published openly. Fractional GPU options are unusual and useful. | Hourly, published, with a strong simplicity pitch. |
| Getting started | Credit card. | Credit card. |
| Capacity | Generally available; broad geographic spread is the differentiator. | Modest. |
| Ownership | Private, long-established as a general cloud before GPUs. | Private. |
Vultr
For: Teams that want a GPU near their users rather than near Virginia, and anyone who needs a fraction of a GPU.
The catch: Not the cheapest and not the biggest. It wins on geography and on fractional sizes, which are the two things bigger providers handle badly.
Economics: Fractional GPU pricing makes small inference workloads viable that would be absurd on a whole H100.
Civo
For: Kubernetes-native teams who found the hyperscaler console exhausting.
The catch: Small catalogue of parts. If you need the newest silicon this is not where it lives.
Economics: Competitive on the parts it carries, and the pricing is genuinely simple to reason about — rarer than it should be.
Neither table row is a price
Deliberately. Published on-demand rates at this layer move weekly, and essentially nobody signing a real contract pays them — every serious buyer pays less than every list figure either of these companies publishes. Quoting one here would date this page within a month.
The GPU rental price index carries dated, sourced figures instead, and the durable finding there is the spread: the identical H100 rents from roughly $1.38 to $12.29 an hour depending only on who you rent it from.
The layers underneath both
Whichever you pick is renting you chips in a building that needs power. In 2026 that is the constraint that binds: Microsoft has disclosed an Azure backlog it cannot fill for want of megawatts rather than accelerators, and the US interconnection queue exceeds 2,600 GW with roughly 80% of projects withdrawing before they energise.
Layer 2 — Silicon · Layer 1 — Energy · The interconnection queue