CoreWeave is a pure-play NVIDIA GPU cloud whose entire differentiation lives at three layers — the compute fabric (0), infrastructure orchestration (2A), and application runtime (2B) — and which is, unusually for a cloud in this assessment, built deliberately on open substrate. Its orchestration opinions rest on Kubernetes (CKS) and Slurm (SUNK), not a proprietary managed scheduler, and SUNK Anywhere runs those same workflows on non-CoreWeave and on-prem clusters with few configuration changes. That makes CoreWeave's orchestration layer the rare cloud cell that reads as Retained rather than Ceded.
The capture mechanism is decoupled and invisible. The substrate the buyer sees — Kubernetes, Slurm, an S3-compatible object store with no egress fees — is genuinely open and portable, which is reassuring at purchase. The value that actually accumulates sits in two captive places the buyer underprices: the silicon and fabric (wholly NVIDIA, with no x86-OEM substitutability because you rent CoreWeave's fleet rather than buy swappable hardware), and the Weights & Biases opinion layer at the top of the stack — experiment history, evaluation frameworks, and the closed training-to-inference 'Superintelligence Loop' — which is proprietary and does not move when the substrate does.
The data layers are thin. CoreWeave AI Object Storage is a performance substrate (LOTA caching, InfiniBand, 7 GB/s per GPU, cross-cloud reach), not a governed data foundation: there is no catalog, lineage, classification, or policy engine comparable to AWS Lake Formation, Google Knowledge Catalog, or VAST Catalog, and no native vector/RAG or pipeline product. Layers 1A, 1B, and 1C are gaps the enterprise must fill with its own open tooling — which keeps authority Retained there by default, but is a finding, not a strength.
The buyer's trade: CoreWeave offers the latest NVIDIA silicon, best-in-class GPU-cluster orchestration on open standards, and a credible agentic development loop via W&B — with a far more favorable orchestration-layer authority profile than any hyperscaler. In exchange it cedes total dependence on NVIDIA at Layer 0 (supplier, strategic partner, and equity holder after the January 2026 $2B investment) and accumulates captive value in the W&B value plane. The instrument's reading: open where it is cheap to be open, captive where the value compounds.
Layer-by-layer status: Layer 0 (Ceded to NVIDIA), Layer 1A (Performance substrate, not governance), Layer 1B (Enterprise-provided), Layer 1C (Bytes move fast, pipelines absent), Layer 2A (Strong — and Retained (open substrate)), Layer 2B (Strong — authority splits by what you use), Layer 2C (Intelligence-2C (improvement loop), Infrastructure-2C absent), Layer 3 (+1) (W&B developer plane — captive opinion layer).
Assessment framework: 4+1 Layer AI Infrastructure Model. Scoring model: Decision Authority Placement Model (DAPM) — Retained, Delegated, or Ceded. Published by The CTO Advisor LLC. Author: Keith Townsend. Date assessed: May 29, 2026. Version: v1.0.
Raw compute, networking, and acceleration fabric
Single-GPU through 8× NVLink systems and multi-node InfiniBand clusters; HGX B300 with 2.1 TB HBM3e; GB200/GB300 rack-scale; Vera Rubin NVL72 expected H2 2026. Rented capacity, not owned hardware.
NVIDIA Quantum-X800 InfiniBand, ConnectX, BlueField DPUs; multi-cloud backbone with private interconnects, direct cloud peering, 400 Gbps-capable ports. No commodity-substrate equivalent.
First-to-market NVIDIA generations: HGX B300 GA, GB200/GB300 rack-scale, Hopper/Ada; among first to deploy Vera Rubin NVL72 in H2 2026.
Quantum-X800 InfiniBand, ConnectX NICs, BlueField DPUs for node/resource offload — the entire fabric is NVIDIA silicon.
CoreWeave's Layer 0 is differentiated and complete — the freshest NVIDIA fleet of any cloud, 40+ data centers, ultra-low-latency fiber, BlueField-offloaded bare-metal nodes, and MLPerf-leading utilization. Calibrates with AWS and NVIDIA at this layer: strong capability, zero buyer authority over silicon or fabric. The authority call is harder than it looks. Dell and HPE score Retained at Layer 0 because x86/NVIDIA hardware is substitutable across OEMs — the enterprise owns the box and can swap the vendor. CoreWeave is not that: the enterprise rents CoreWeave's NVIDIA-only fleet, so there is no commodity substrate it controls and no swap that does not mean leaving CoreWeave. The dependence is also uniquely total — NVIDIA is supplier, Elite Cloud Partner, and (after the January 2026 $2B investment) a significant equity holder. There is no alternative-accelerator path here as there is on AWS (Trainium) or Google (TPU). Ceded.
The NVIDIA column is the densest in this row. Unlike the hyperscalers, CoreWeave has no own-silicon hedge — NVIDIA dependence at Layer 0 is the structural fact of the business.
Durable, governed data foundation — the Governance Catalog that Layer 2C queries
Assessment pending
CoreWeave AI Object Storage (powered by LOTA — Local Object Transport Accelerator) is a genuinely strong storage product: S3-compatible, fully managed, distributed GPU-local cache, up to 7 GB/s per GPU, cross-region and cross-cloud single-dataset access, no egress or request fees, and >75% lower cost than typical alternatives. But it is a performance substrate, not a governed data foundation. There is no catalog, no automated classification, no lineage, no compliance tagging, and no policy enforcement engine — nothing comparable to AWS Lake Formation + Glue, Google Knowledge Catalog, or VAST Catalog. The governance function the 4+1 model defines at this layer remains entirely the enterprise's responsibility. That makes this a gap, not a strength — and gap, by the methodology, means the DAPM is Retained by default because no vendor has claimed the function. Note the decoy: 'your data is in an open S3-API store, no egress, runs anywhere' is true and reassuring, but says nothing about governance authority. The storage bytes were always the cheap-to-rebuild part. The governed-artifact layer that would create capture simply does not exist here.
If a buyer treats AI Object Storage as their governance layer, they are mistaking throughput for governance. The instrument scores the function, which is absent.
Vector/hybrid search and RAG context assembly
Assessment pending
No native vector database, hybrid search, or managed RAG context service. W&B Weave provides tracing and evaluation for agent and LLM behavior, not retrieval. Tensorizer accelerates model loading, not context assembly. The enterprise brings its own retrieval stack (pgvector, Milvus, Weaviate, etc.) and runs it on CoreWeave compute. Calibrates with NVIDIA's 1B gap: CoreWeave accelerates retrieval workloads at Layer 0/2B but does not provide the retrieval capability itself. Function remains the enterprise's — gap, DAPM Retained.
Confirmed by absence in current product set, not inferred from a single doc.
ETL/ELT, transformation, lineage, cost-aware data movement
Assessment pending
LOTA moves bytes fast and across clouds, and AI Object Storage gives a single dataset global reach without replication. But fast data transport is not a data pipeline. There is no ETL/ELT, no transformation framework, no lineage tracking, and no cost-aware movement orchestration product. The enterprise composes this from open tooling (Airflow, Ray Data, dlt, Spark) on CoreWeave compute. Data movement ≠ data pipelines. Cross-cloud reach is a Layer 0/1A throughput property, not a Layer 1C pipeline capability. Function remains the enterprise's — gap, DAPM Retained.
GPU scheduling, quotas, fair-share, topology-aware placement
Managed Kubernetes with Slurm integrated as a K8s scheduler; login/compute/controller nodes as Pods; preemption across Slurm and K8s workloads. Built on CNCF Kubernetes and open-source Slurm. SUNK Anywhere extends to non-CoreWeave and on-prem clusters.
Proprietary cluster health, straggler detection, silent-fault mitigation, and node lifecycle management; ~50% fewer interruptions claimed. Operational intelligence layer captive to CoreWeave.
SUNK topology-aware scheduling tuned to GB200 rack-scale NVLink/InfiniBand layouts; Mission Control detects GPU stragglers and silent hardware faults.
This is CoreWeave's signature layer and the most consequential authority call in the row. CKS (CoreWeave Kubernetes Service) plus SUNK (Slurm on Kubernetes) plus Mission Control deliver differentiated, complete GPU-cluster orchestration: unified Slurm batch scheduling and Kubernetes container orchestration on one cluster, topology-aware placement, preemption logic across both schedulers, automated node lifecycle and straggler mitigation, and self-service cluster provisioning. Strong capability — peer to the hyperscalers' managed orchestration and, by several customer accounts, ahead of it for large training clusters. The authority reading is where CoreWeave diverges from every hyperscaler in this assessment. The hyperscalers' managed orchestration is scored present-and-Ceded because the scheduling opinions (Karpenter, proprietary fair-share) are captive. CoreWeave's orchestration opinions rest on open substrate — Kubernetes (CNCF) and Slurm (open-source, SchedMD) — and SUNK Anywhere explicitly runs the same workflows on non-CoreWeave clusters and on-prem 'with very few configuration changes,' confirmed by customers running it across providers. The enterprise can lift its Slurm/K8s scheduling opinions and operate them elsewhere without rebuilding. By the litmus, that is Retained, and it calibrates near the IBM/Red Hat standard. Two proprietary slivers sit inside the otherwise-Retained layer: Mission Control (the health/straggler-detection and lifecycle service) and the SUNK self-service operator/console are CoreWeave IP. A buyer who depends on Mission Control's operational intelligence specifically inherits a Ceded dependency — but the core scheduling abstraction they would carry to another platform is open.
Pressure-tested against the tidy-story risk: it would be neat to call every cloud's 2A Ceded. CoreWeave genuinely breaks that pattern because it ships the open schedulers as the product rather than hiding a proprietary one behind a managed service. The Retained call is evidence-driven (SUNK Anywhere portability), not generous.
Model serving, training execution, agent runtime
Open model-serving on managed Kubernetes; Tensorizer accelerates model load from S3. Runtime opinions portable off CoreWeave.
Always-on production inference with integrated monitoring; environment-free Serverless RL for agentic post-training (~40% cost reduction, ~1.4× faster claimed). Proprietary runtime.
Inference and training tuned to NVIDIA GPUs; integrates NVIDIA stack but does not require NVIDIA AI Enterprise as the only runtime path.
Strong, complete runtime capability across the training-to-inference span: CKS natively integrates open serving stacks (KServe, KubeFlow), Tensorizer streams serialized models from S3/HTTPS for fast cold-start (5× faster downloads claimed), and CoreWeave Inference runs continuously-on production workloads with built-in scaling and health monitoring. W&B Serverless RL adds post-training (RL) execution without provisioning. Authority is genuinely split and depends on the presented path the buyer chooses — flagged here rather than averaged away: • Open serving (KServe/KubeFlow on CKS) — the runtime opinions are open and portable. Retained. • CoreWeave Inference / W&B Serverless RL — proprietary CoreWeave/W&B runtime; adopting it means the execution opinions are captive. Ceded where used. Calibrates with NVIDIA 2B (NIM/Dynamo runtime, strong/Ceded) on the proprietary path, but CoreWeave — unlike NVIDIA — also offers a fully open serving path that keeps the layer Retained. Strong capability, mixed authority.
This is the one cell resting partly on inference: the DAPM depends on which runtime a given customer adopts. Scored as presented architecture (both paths shipped), with the split named rather than collapsed.
Policy-driven placement and coordination of agents and inference
Closed training-to-inference loop: production traces feed RL post-training; Weave evaluates and monitors agent behavior; Skills/MCP drive autonomous improvement. Intelligence-2C (which agent improves, under which evaluation). Proprietary W&B/CoreWeave IP.
Assessment pending
CoreWeave's May 2026 'Superintelligence Loop' — Serverless RL + CoreWeave Inference + W&B Weave observability + W&B Skills — is real Intelligence-2C: a closed feedback loop that governs which agents improve, evaluates agent actions against custom signals, surfaces failure modes in multi-agent workflows, and (via Skills + MCP server) turns coding agents into autonomous agent-improvers. It is productized and adopted, not a slideware claim. But routing is not reasoning, and improvement is not placement. This is closed-loop agent operations and continuous improvement, not live Infrastructure-2C: no policy engine answers 'given data residency, cost, latency, and compliance, run this inference on B300 in region X versus region Y at request time.' The physical placement underneath is just SUNK/Kubernetes scheduling. Calibrates with AWS — which has productized Intelligence-2C (AgentCore Policy/Guardrails) and absent Infrastructure-2C — but CoreWeave's Intelligence-2C is narrower: it governs agent improvement and evaluation, not a general action-authorization policy plane like Cedar-based AgentCore Policy. So moderate, not strong. The live per-inference placement gap is universal across this assessment, not specific to CoreWeave; it is noted, not penalized as a unique defect. Authority: the W&B improvement/observability loop is proprietary IP — Ceded where adopted. The underlying scheduling is the Retained SUNK/K8s layer.
Pressure-tested 'routing is not reasoning': the Superintelligence Loop is genuinely agentic but operates on the model/agent-quality axis, not the request-placement axis. Scored on the same basis as the other rows.
AI-powered business capabilities, developer and MLOps workflows
Experiment tracking, model versioning, artifact lineage, lifecycle promotion. The accumulated experiment and lineage history is the captive value.
Agent/LLM tracing, evaluation framework, production monitoring, Playground; Skills + MCP server for autonomous agent improvement. Proprietary developer value plane.
Assessment pending
Through Weights & Biases (acquired 2025), CoreWeave owns a genuine, widely-adopted value plane: W&B Models + Registry (experiment tracking, model versioning, lineage, lifecycle promotion), W&B Weave (tracing, evaluation, production monitoring, Playground), and W&B Skills. W&B powers 1,500+ organizations including 30+ foundation-model builders. This is a real Layer 3 opinion the vendor owns and operates, not merely an ISV catalog — so it is not scored 'partner.' It is narrower than the hyperscaler value planes, which is why it is moderate rather than strong. W&B is a developer/MLOps and agent-ops surface, not a breadth of business-workflow applications comparable to AWS Bedrock/Q/Kiro or Google's application stack. There is no business-process application ecosystem here — the value plane is for the people who build AI, not the people who consume it in line-of-business workflows. This is the decoupled capture, and it is the most important reading in the row. The substrate the buyer chose CoreWeave for — open Kubernetes, open Slurm, S3-API storage, no egress — is reassuringly portable. The value that actually compounds — years of experiment history, evaluation frameworks, the improvement loop, the model registry's lineage — accumulates inside W&B, which is proprietary with no open exit. The buyer feels free because the visible layer is open, and underprices the W&B commitment precisely because everything beneath it can move. Ceded.
The W&B capture is the inverse of the hyperscalers' coupled capture: there the data lives visibly in the vendor's namespace; here the data stays open while the opinion layer above it is captive. More reassuring at purchase, which is what makes it worth naming.
CoreWeave is a pure-play NVIDIA GPU cloud whose entire differentiation lives at three layers — the compute fabric (0), infrastructure orchestration (2A), and application runtime (2B) — and which is, unusually for a cloud in this assessment, built deliberately on open substrate. Its orchestration opinions rest on Kubernetes (CKS) and Slurm (SUNK), not a proprietary managed scheduler, and SUNK Anywhere runs those same workflows on non-CoreWeave and on-prem clusters with few configuration changes. That makes CoreWeave's orchestration layer the rare cloud cell that reads as Retained rather than Ceded.
The capture mechanism is decoupled and invisible. The substrate the buyer sees — Kubernetes, Slurm, an S3-compatible object store with no egress fees — is genuinely open and portable, which is reassuring at purchase. The value that actually accumulates sits in two captive places the buyer underprices: the silicon and fabric (wholly NVIDIA, with no x86-OEM substitutability because you rent CoreWeave's fleet rather than buy swappable hardware), and the Weights & Biases opinion layer at the top of the stack — experiment history, evaluation frameworks, and the closed training-to-inference 'Superintelligence Loop' — which is proprietary and does not move when the substrate does.
The data layers are thin. CoreWeave AI Object Storage is a performance substrate (LOTA caching, InfiniBand, 7 GB/s per GPU, cross-cloud reach), not a governed data foundation: there is no catalog, lineage, classification, or policy engine comparable to AWS Lake Formation, Google Knowledge Catalog, or VAST Catalog, and no native vector/RAG or pipeline product. Layers 1A, 1B, and 1C are gaps the enterprise must fill with its own open tooling — which keeps authority Retained there by default, but is a finding, not a strength.
The buyer's trade: CoreWeave offers the latest NVIDIA silicon, best-in-class GPU-cluster orchestration on open standards, and a credible agentic development loop via W&B — with a far more favorable orchestration-layer authority profile than any hyperscaler. In exchange it cedes total dependence on NVIDIA at Layer 0 (supplier, strategic partner, and equity holder after the January 2026 $2B investment) and accumulates captive value in the W&B value plane. The instrument's reading: open where it is cheap to be open, captive where the value compounds.