CoreWeave Named Only Platinum AI Cloud in SemiAnalysis ClusterMAX Rating
CoreWeave says it is the only AI cloud provider to receive the top Platinum rating in SemiAnalysis's ClusterMAX Rating System. The company published the news on its blog, alongside a summary of the methodology and its own account of how it builds its platform.
- SemiAnalysis's ClusterMAX system rates GPU clouds in five tiers, from Platinum to UnderPerform.
- CoreWeave says it is the only AI cloud provider at the Platinum tier.
- SemiAnalysis cited CoreWeave's reliable operation of 10k+ H100 clusters.
- CoreWeave claims 20% higher GPU cluster performance and 96% goodput versus a 90% industry average.
- SemiAnalysis states that the report was prepared independently.
What SemiAnalysis rated and how
According to CoreWeave, SemiAnalysis evaluated dozens of providers, including AWS, GCP, Azure, Crusoe, Nebius and Lambda, from the perspective of an average reasonable customer. Over the last 12 months, SemiAnalysis independently tested and/or collected customer feedback from dozens of cloud providers it considered ready for AI workloads. It then built the GPU Cloud ClusterMAX Rating System.
CoreWeave describes the testing as extensive benchmarking and profiling meant to represent customer needs. SemiAnalysis also tried to scale workloads from single nodes to thousands of GPUs. Providers were assessed across several dimensions, including security, performance, reliability, technical expertise and operational posture.
The system has five tiers, from highest to lowest: Platinum, Gold, Silver, Bronze and UnderPerform. CoreWeave says that, according to SemiAnalysis, it is currently the only AI cloud that is "raising the industry bar and qualifies for ClusterMAX Platinum".
What the report said about CoreWeave
SemiAnalysis also recognized CoreWeave's industry leadership in reliably operating large-scale 10k+ H100 clusters, the company says. In the report, SemiAnalysis concluded that "CoreWeave is clearly leading in providing the best GPU cloud experience and has very high goodput". It added that the company is entrusted with large-scale GPU infrastructure for AGI labs like OpenAI, high frequency trading firms like Jane Street, and even NVIDIA's internal clusters.
Dylan Patel, Chief Analyst at SemiAnalysis, pointed to CoreWeave's cluster burn-in and automated node lifecycle controller. These run passive and active health checks to catch unhealthy nodes and identify issues such as link flaps, convergence issues and silent data corruptions.
CoreWeave notes that the ClusterMAX system and the report were prepared independently by SemiAnalysis. It also states that no part of SemiAnalysis's compensation from its clients was, is, or will be directly or indirectly related to the specific tiering, ratings or comments in the article.
CoreWeave's own performance claims
The company uses the announcement to restate figures about its platform. These are CoreWeave's claims, not SemiAnalysis findings. According to CoreWeave, performance optimizations across infrastructure, CoreWeave Kubernetes Service and Slurm on Kubernetes (SUNK) deliver 20% higher GPU cluster performance than alternative solutions. It also says its focus on reliability means 50% fewer interruptions and higher system goodput, at 96% versus a 90% industry average.
CoreWeave adds that everything is included in the price, with no hidden fees or surprise support costs. It names OpenAI, Mistral, Cohere, IBM and Jane Street as customers that choose it for AI workloads.
Security and isolation
Security is one of the dimensions the blog covers in detail. CoreWeave says it uses Extended Detection and Response (XDR) and Data Loss Prevention (DLP), and adheres to SOC2 and ISO 27001. It also employs in-house information security teams.
CoreWeave Kubernetes Service runs on bare metal instances where every node is single-tenant. The company says this mitigates container escape vulnerabilities, a common risk with traditional hypervisor based approaches. It also lists encryption, identity management, tenant networking isolation on both Ethernet and InfiniBand, VPC and Direct Connect networking, and proactive patching of CVEs. For customers with sensitive data, it offers dedicated data center deployments.
What it means for customers
For buyers comparing GPU clouds, the ClusterMAX tiers offer a third-party reference point, with the caveat that CoreWeave is the party promoting this result. The Platinum tier, as SemiAnalysis defines it, covers security, price for value, technical expertise, reliability backed by clearly defined SLAs, managed Slurm and Kubernetes offerings, and networking performance. The full SemiAnalysis report is published separately for those who want to check the criteria and the ratings of other providers.
