CoreWeave and Nebius Earned Top ClusterMAX Ratings

The latest evaluation of cloud infrastructure providers highlights performance leaders in the AI training market.

Updated on Sept. 23, 2026 in Data Centers

Bold flat-color editorial illustration of nested geometric server chassis arrays, representing high-performance cloud infrastructure assessment.
CoreWeave and Nebius received Platinum ClusterMAX ratings in the latest 3.0 cycle assessment, which evaluated over 300 cloud infrastructure providers. AI Illustration. Upload story photo >

Live Poll

Do you trust specialized 'neocloud' providers to handle critical AI infrastructure better than hyperscale giants?

CoreWeave and Nebius have secured the Platinum ClusterMAX rating in the latest 3.0 cycle assessment from SemiAnalysis. This report, which evaluated 323 cloud providers, identified these firms as top performers in infrastructure benchmarks.

Why it matters

These ratings provide critical transparency in the competitive AI cloud market, identifying which providers can reliably support large-scale computing tasks. The evaluation highlights current leaders in the field as demand for high-performance training clusters continues to grow.

CoreWeave maintained a 96% goodput rating, which measures the efficiency of successfully completed tasks versus total workload, outperforming the industry average of 90%. The company currently operates over 10,000 H100 clusters.

The players

CoreWeave

A specialized cloud infrastructure provider focused on high-performance computing and large-scale AI cluster deployments.

Nebius

An AI infrastructure provider that achieved the Platinum ClusterMAX rating for the first time in the 3.0 cycle.

SemiAnalysis

An industry research and consulting firm that publishes technical evaluations of semiconductor and cloud infrastructure performance.

The details

The ClusterMAX 3.0 report assessed 77 cloud providers in depth using a methodology that incorporated over 200 end-user interviews. Evaluators examined ten distinct categories including security protocols, workload orchestration, storage throughput, and network architecture. Goodput serves as a key metric to identify how effectively a provider's infrastructure translates raw hardware capacity into successful computational output.

Timeline

  1. March 2025: The benchmark launched and the first report was issued.

  2. November 2025: The ClusterMAX 2.0 evaluation cycle took place.

  3. September 23, 2026: The ClusterMAX 3.0 report was released.

The Tech Race

The ClusterMAX report functions as a standardized benchmark for AI infrastructure, similar to how independent labs provide energy efficiency ratings for consumer electronics. The move from CoreWeave holding sole Platinum status to the inclusion of Nebius indicates a broadening field of competitors in high-performance cloud computing.

Users seeking reliable AI training infrastructure can use these rankings to filter cloud providers based on verified goodput performance. Businesses relying on massive GPU clusters may use these findings to evaluate potential vendors for their next training cycle or hardware procurement.

The takeaway

The addition of new providers to the Platinum tier suggests that the capability to support large-scale H100 clusters is becoming more broadly available. Readers should monitor future ClusterMAX updates to see if other providers reach the 96% goodput milestone or if performance gaps continue to shift.

Further reading

For more on the current state of infrastructure for high-performance computing, see our Data Centers section.

Live Poll

Do you trust specialized 'neocloud' providers to handle critical AI infrastructure better than hyperscale giants?

CoreWeave and Nebius Earned Top ClusterMAX Ratings