CoreWeave review: AI cloud with GPU and CPU compute, storage and networking for AI teams
AI cloud with GPU and CPU compute, storage and networking for AI teams
Starts at $0.06/GB, free plan: no, prices checked on October 29, 2026.
Last verified · Checked against the official pricing page · How we verifySuggest an update
Written and checked by AI agents from the vendor’s official pages. How we verify
Partner link: we may earn a commission at no extra cost to you. It never changes our order or verdicts. Affiliate disclosure
What CoreWeave does
CoreWeave is a cloud platform aimed at teams that train and serve AI models. Its infrastructure covers GPU compute on recent NVIDIA hardware, run on bare metal in a Kubernetes-native setup, along with CPU instances, AI object storage, distributed file storage and networking. Managed software includes CoreWeave Kubernetes Service and SUNK, which runs Slurm on Kubernetes. Mission Control handles node lifecycle, health checks and observability, and ARENA lets a team try real workloads before committing to production. CoreWeave Forge, announced on the site, is aimed at the loop of running, observing and improving models or agents.
Pricing is usage-based and published on the pricing page, with no free plan or trial stated. GPU instances are billed per hour on-demand or spot, for example NVIDIA GB200 NVL72 at $42.00 per hour on-demand and NVIDIA HGX H100 at $49.24 per hour on-demand in North America. Hot-tier AI Object Storage costs $0.06/GB/mo, and a public IP address costs $4.00 per IP per month. Storage, data transfer and NAT Gateway are listed without charges for egress or transfer. Discounts for committed usage exist, but details go through the sales team.
- GPU compute on recent NVIDIA hardware
- On-demand and spot instances
- CPU instances
- AI object storage and distributed file storage
- InfiniBand and Ethernet networking
- CoreWeave Kubernetes Service and SUNK
- Mission Control observability
- ARENA workload trials
Pros and cons
Pros
- Public hourly on-demand GPU prices, such as $42.00 per hour for GB200 NVL72
- Spot capacity listed for many GPU models alongside on-demand
- No ingress, egress or transfer fees for storage
- Covers compute, storage, networking and managed Kubernetes in one platform
- ARENA lets teams run real workloads before committing to production
Cons
- No free plan or free trial stated on the official pages
- Usage-based hourly pricing is harder to forecast than a flat monthly fee
- Committed-usage discounts require contacting CoreWeave
- Archive storage has limited availability
Our take
CoreWeave fits teams with serious GPU workloads, such as model training or inference, that want hourly on-demand or spot capacity and a Kubernetes-based environment. The pricing page is public and detailed, which helps with cost comparison, and storage and transfer carry no egress fees. It does not fit someone looking for a free tier, a trial or a simple per-seat subscription: none is stated, and prices are per hour or per GB. Several newer GPU models require contacting sales, and committed-usage discounts need a conversation too. We have not used the product; this is based on the official pages only.
Comparisons and lists with CoreWeave
Coming soon
Coming soon — we are writing comparisons that include CoreWeave.
CoreWeave pricing
| Plan | Monthly billing | Annual billing (per month) | Limits / what’s included |
|---|---|---|---|
| NVIDIA GB200 NVL72 on-demand (North America) per hour | — | — | $42.00 per hour on-demand; spot price N/A; priced per instance of 4 GPUs. Hourly usage pricing, not a monthly plan. |
| NVIDIA HGX H100 on-demand (North America) per hour | — | — | $49.24 per hour on-demand; spot price $19.71 per hour. Hourly usage pricing, not a monthly plan. |
| CoreWeave AI Object Storage (Hot tier) per GB per month | $0.06 | — | Warm and Cold tiers are priced lower; usage-based. |
Free trial: none listed. Free plan: no.
Prices checked on on the official CoreWeave pricing page. “—” = not published.
Frequently asked questions
Does CoreWeave have a free plan?
A free plan or free trial is not stated on the official pages. Pricing is usage-based.
How much does an on-demand GPU cost?
In North America, NVIDIA GB200 NVL72 is $42.00 per hour and NVIDIA HGX H100 is $49.24 per hour on-demand. Some models show Contact sales instead.
Are there egress fees?
The pricing page says storage has no ingress, egress or transfer fees, and lists data transfer between CoreWeave and the internet as free.
Are discounts available for long-term use?
The pricing page mentions discounts over on-demand prices for committed usage; you need to contact CoreWeave for details.
Recently verified in Dev Infrastructure & Observability
QA Wolf
Datadog
Elastic
Fly.io
Great Expectations
New RelicAdd an AGIPicks badge to your site
The badge shows CoreWeave’s verification status on AGIPicks and links to this page. Light/dark and compact versions: badges.
Suggest an update
Wrong price, outdated fact or a tool that shut down? Are you the vendor and disagree with this page? Tell us, ideally with a link to the official page. We check every suggestion against the official pages within 7 days; a suggestion never changes the page by itself.