Most AI teams operating at scale run across multiple clouds. Their stack, assembled over the years from many vendors for everything from training and inference to security and storage, grows harder to manage with every new project. Each time a team expands onto a new cloud, connecting that stack to new infrastructure becomes the slowest and riskiest part of the move, and the platform team is left to verify that every integration actually works. Until now, the partner collaborations that could ease that work ran one relationship at a time, largely invisible to customers.
Today, we're introducing the CoreWeave Partner Network: a curated network of technology partners offering integrations that are validated on CoreWeave under real-world conditions. It brings together the valuable relationships CoreWeave has built over years—with CrowdStrike, VAST Data, Reflection, IBM, and many more—into one place, so your team can see exactly which tools are proven and ready to run on CoreWeave Cloud and rapidly adopt them with confidence.
How the CoreWeave Partner Network works
The CoreWeave Partner Network is a proven network of technology partners spanning four segments that together cover the full AI stack:
- Infrastructure: Hardware and systems partners covering compute, storage, networking, servers, and interconnects, including VAST Data, Dell, AMD, IBM, and Pulumi
- Data services: Partners that improve how AI data is stored, processed, moved, and accessed across the AI lifecycle, including ClickHouse, Encord, LanceDB, and Datology
- ISVs: Software partners—including Grafana, Datadog, Union, SkyPilot, IBM, Pinecone, Redis, LlamaIndex, and Clockwork—spanning orchestration, observability, and model development, with a dedicated security segment anchored by CrowdStrike, Zscaler, and Cranium
- Models and inference: Partners that provide the foundational software and services used to build, train, serve, and scale AI applications, including Reflection, RadixArk, Inferact, OpenRouter, and Red Hat
Together, they support every stage of the AI loop, from running and observing models in production to curating data, improving, and evaluating what ships next.
Every partner in the CoreWeave Partner Network meets the same bar: a working integration, tested and proven on CoreWeave infrastructure under real-world production conditions. Partners join based on real demand; they're in the network because our customers already run them or are asking for them.
Partner technology proven at production scale
Customers are already building on CoreWeave with partner technology throughout their stack:
- IBM trained its Granite models on CoreWeave, built around one of the first instances in the world to stand up NVIDIA GB200 NVL72 clusters
- Cohere built North, its agentic AI platform, on one of the industry's first production NVIDIA GB200 NVL72 clusters, with 3x faster training compared to NVIDIA Hopper
- Mistral AI trains its models entirely on CoreWeave, citing 2.5x faster training on NVIDIA GB200 NVL72s
- Decart runs NVIDIA HGX B200 at production scale through its collaboration with CoreWeave, whose platform includes Weights & Biases.
The CoreWeave Partner Network gives every team a proven path to similar results.
Built on years of collaborations
The CoreWeave Partner Network formalizes relationships that go back years. CoreWeave and its AI developer platform, Weights & Biases, have run production integrations with Encord, Pinecone, and others from the first wave of open-source LLMs through to the AI labs training frontier models today.
These collaborations weren't built for this announcement—some go back years, one customer relationship at a time. What changes today is visibility: the depth we've always had with partners like these is now something a customer can actually see before they build. Others are here because they aligned with our vision of where the market was heading.
Jon Jones, Chief Revenue Officer, CoreWeave
VAST Data, CrowdStrike, and IBM have been part of that history from the start.
Compute and data have to operate as one system. By combining CoreWeave's purpose-built AI cloud with the VAST AI Operating System, we give AI builders an architecture designed to evolve with them, from large-scale training and fine-tuning, to production inference and increasingly continuous, agentic workloads.
Alon Horev, Co-Founder and CTO, VAST Data
CoreWeave powers some of the most advanced AI workloads on the planet, and CrowdStrike secures where AI happens. Together, we're building a foundation for the agentic era, where security is inherent to AI itself.
George Kurtz, CEO and Founder, CrowdStrike
What's exciting about this collaboration is the opportunity to help clients put AI to work quickly. By combining CoreWeave's purpose-built AI cloud with IBM and Red Hat technologies, we're giving enterprises more choice, addressing integration complexity, and creating a path to deploy and scale AI workloads. The result is that clients can spend less time stitching together infrastructure and more time driving innovation to realize business value from AI.
Adam Lawrence, General Manager, IBM Americas
Until now, these collaborations operated one relationship at a time, largely invisible to customers. The CoreWeave Partner Network makes them accessible in one place and creates a repeatable motion for what comes next: deeper integrations, joint go-to-market, and shared innovation with the vendors that AI teams already depend on.
The infrastructure was already here. The tooling was already here. Now the partners that help you run every stage of the loop are here too.
What this means for your team
Bring the stack you already trust
The tools your team runs today—serving layers, observability dashboards, security tooling, and more—are already integrated on CoreWeave Cloud. Expanding onto CoreWeave for scale doesn't mean rebuilding what already works. Inference partners keep your models serving without a rebuild; Grafana and Datadog carry over the observability you've built runbooks around; and CrowdStrike, Zscaler, and Cranium anchor a security segment so your existing threat visibility and response tooling comes with you.
Start from integrations that already work
Integrations that pass on a spec sheet can still fail under real traffic. Every integration in the CoreWeave Partner Network is tested directly on CoreWeave infrastructure under production conditions before it reaches you, so the validation work is done before you ever touch it.
Skip the sorting work
A marketplace with thousands of listings hands the vetting over to your team. Asking an LLM gets you the market's consensus or what's popular, not what's proven. The CoreWeave Partner Network is deliberately curated: select partners join when customers are already running them or asking for them, and every segment is held to the same standard. The result is a curated, vetted list that still spans every layer of your stack.
Build on anything, locked-in to nothing
The CoreWeave Partner Network will continue to expand. The same demand-led model that brought infrastructure, data, and security partners into the network is extending to the tools your agents call, too, starting with search. Exa, Parallel Web Systems, and You.com join as partners to give agents built on CoreWeave a shared search layer, reachable through one integration instead of a separate contract for each.
Most closed models come with web search built in. Most open-weight models don't, so teams choosing open weights for cost, control, or customization have had to bolt search on themselves, one provider integration at a time.
That's the gap this closes. Turn it on with a single setting on a standard inference request, or reach it the same way from inside a CoreWeave Sandbox; no separate SDK, no separate account to stand up. And, because this isn't an exclusive arrangement, choosing this integration today doesn't lock in what's available tomorrow. We're announcing this at Fully Connected, with the rest of the integration following in the weeks after.
This collaboration follows the same principle behind everything else in the CoreWeave Partner Network: proven solutions, not a list assembled for an announcement. As these integrations come online, they'll be held to the same bar as every partner here today.
Get started with the CoreWeave Partner Network
Explore the CoreWeave Partner Network to see which partners are validated and ready to run on CoreWeave. If you're at Fully Connected 2026 this week, visit our partner sessions to see the integrations in action.
Want to go deeper on a specific integration? Talk to your CoreWeave account team about your workloads.
