Prices updated 2026-06-23 — sourced from live provider data

CoreWeave MI300X Pricing Options

All available MI300X instance types on CoreWeave, sorted by lowest hourly rate.

Region Instance Type Price/hr GPUs vCPUs RAM
US MI300X 192GB On-Demand $2.5000/hr Cheapest 1× GPU 24 vCPU 320 GB

About CoreWeave MI300X Cloud Pricing

CoreWeave is a GPU-specialized cloud provider built from the ground up for compute-intensive workloads. Unlike hyperscalers, CoreWeave's entire infrastructure is optimized for GPU density with bare-metal performance and Kubernetes-native orchestration.

Why Choose CoreWeave for MI300X

  • Purpose-built GPU infrastructure (not repurposed general cloud)
  • Kubernetes-native with built-in autoscaling
  • NVLink and NVSwitch interconnect for multi-GPU training
  • Typically 30–50% cheaper than hyperscalers for equivalent GPU compute

Best suited for: AI startups and research labs that need raw GPU performance without paying the hyperscaler tax. Especially strong for large-scale training jobs and inference at scale.

MI300X Technical Overview

ArchitectureAMD CDNA 3
Memory192 GB HBM3
Best ForLarge model training and inference for teams building on the AMD ROCm ecosystem

CoreWeave deploys H100 SXM5 in purpose-built GPU pods with NVLink/NVSwitch interconnects enabling 3.35 TB/s GPU-to-GPU bandwidth -- essential for distributed training across multiple H100s. Unlike Lambda, CoreWeave offers both on-demand and reserved 1-year contracts (typically 25-35% savings for reserved). CoreWeave's Kubernetes-native orchestration (EKS-compatible) allows auto-scaling training clusters with GPU instance pre-warming. Their H100 clusters run on custom liquid-cooled infrastructure, which reduces thermal throttling during extended training jobs. For foundation model pre-training or RLHF at scale, CoreWeave's interconnect topology is a meaningful advantage over providers using PCIe-only H100 configurations.

Pricing Analysis

CoreWeave's cheapest MI300X option starts at $2.5000/hr, which is 20% above the market average of $2.0850/hr. CoreWeave currently offers on-demand pricing only for MI300X. This means no commitment required, but you won't access the deeper discounts available from reserved or spot pricing.

Note: CoreWeave has emerged as a top choice for AI companies that need H100/B200 clusters without the overhead of a full-stack cloud platform.

Market context: MI300X is priced aggressively below equivalent NVIDIA GPUs to compensate for the less mature software ecosystem. Teams that can work with ROCm get more memory and competitive performance for less.

Top Alternatives to CoreWeave for MI300X

Side-by-side cost comparison vs the cheapest CoreWeave MI300X rate ($2.5000/hr).

Provider Price/hr Type vs CoreWeave
RunPod $0.5000/hr On-Demand -80% cheaper vs CoreWeave
Thunder Compute $1.8500/hr On-Demand -26% cheaper vs CoreWeave

View all MI300X providers →

Calculate Your Exact MI300X Cost

Enter your hours, job type, and GPU count. Our calculator compares CoreWeave against all 32 providers instantly.

● $19 — no subscription
Track CoreWeave MI300X prices — daily alerts
Get emailed when CoreWeave drops below your budget. $19 for 30 days. No account needed.
Set Price Alert → Calculator →

Frequently Asked Questions

How much does CoreWeave MI300X cost per hour in 2026? +
CoreWeave MI300X starts at $2.5000/hr (on-demand). The market average for MI300X across all providers is $2.0850/hr. Use the GPU cost calculator to estimate total cost based on your hours and workload type.
What are the cheapest alternatives to CoreWeave for MI300X? +
There are 2 providers offering MI300X cheaper than CoreWeave: RunPod at $0.5000/hr, Thunder Compute at $1.8500/hr. See the full comparison table for all MI300X providers.
Does CoreWeave offer reserved or spot MI300X pricing? +
CoreWeave offers 1 pricing type for MI300X: on-demand. With on-demand only, you pay per hour with no commitment. While you won't get reserved or spot discounts, this provides maximum flexibility for unpredictable workloads.
What workloads is CoreWeave MI300X best suited for? +
The MI300X (AMD CDNA 3 architecture, 192 GB HBM3) is best suited for Large model training and inference for teams building on the AMD ROCm ecosystem. CoreWeave is a gpu native provider — AI startups and research labs that need raw GPU performance without paying the hyperscaler tax. Especially strong for large-scale training jobs and inference at scale.
How much does it cost to run MI300X on CoreWeave for a month? +
At CoreWeave's cheapest MI300X rate of $2.5000/hr: running 24/7 costs ~$60.00/day or ~$1825.00/month per GPU. For training jobs, costs scale linearly — a 100-hour training run would cost ~$250.00. Multi-GPU configurations multiply accordingly. Use the cost calculator for precise estimates with your specific parameters.
Compare across all providers

Not sure which GPU configuration fits your budget?

Get a personalized recommendation based on your monthly budget, workload, and real pricing data from all 58 providers.

Budget Optimizer — $19 →
$19 One-Time

Is a reserved contract worth it here?

Get the break-even math for CoreWeave MI300X — calculated from real 90-day pricing data, not an LLM.

Get Analysis — $19