brannin mcbee

We're coming to India. Home to some of the world's best tech companies and developers. Planning to add 240 MW with @adaniconnex and building a local team. $CRWV

CoreWeave’s full-stack platform turns infrastructure into an AI cloud customers can actually use, expand on, and carry into production. Not just bare metal. A full-stack that integrates orchestration, networking, storage, observability, security, and developer tools to deliver the throughput, reliability, and price-performance customers need.

We have some of the best engineers in the world working alongside some of the most innovative customers to unlock the full potential of AI across scientific discovery, manufacturing, and security. That’s what was on full display at our Fully Connected event this week. $CRWV

We stood up an NVIDIA Vera Rubin NVL72 cluster in days from rack handover to production, delivering it to Cognition to run agentic workloads. The first Vera Rubins powering customer innovation are live on CoreWeave. $CRWV

Lots of big $CRWV news coming out of Fully Connected this week. Some highlights:

At Fully Connected this week. Looking forward to hearing from CoreWeave experts and customers who are building things at scale that couldn’t have been built before now. coreweave.com/fully-connecte… $CRWV

As enterprises put AI agents to work, organizations need clear boundaries around what they can access, which actions they can take, and when human approval is required. Our AI cloud solutions support NVIDIA Open Agent Safety Platform technologies to help customers run agents more securely. We’re building CoreWeave Sandboxes to give agents room to run, with boundaries built in. In-cluster or serverless. $CRWV

We're prepping for our annual customer conference, Fully Connected, which has given me the opportunity to dig into what our sales team is seeing on the ground. Sharing a few observations from the front lines.

>>>Customers Expansion: A frequent trend we see is customers landing with a "smaller" cluster and then rapidly coming back to scale their footprint on CoreWeave Cloud, both on existing and new generations of GPU architecture
-One customer started with test clusters. Less than 36 months after its first contract, its cumulative contract value is up 392x.
-Another customer bought four generations of NVIDIA GPUs from us: A100, H100, H200, GB300. As they upgrade, they keep adding capacity on the older generations too.

>>>Category Expansion: We're seeing rapid diffusion across verticals from financial services and industrial to healthcare and cybersecurity companies
-About 10% of our backlog came from financial services, as of Sep 13, 2026. Quant trading, market making, systematic research. Those firms are moving foundation-model training onto CoreWeave.
-Manufacturers and auto companies are training and simulating world models for physical AI.
-Security companies are deploying large-scale training and low-latency inference for real-time threat detection.

>>>Product Expansion: Customers rely on CoreWeave for our full and growing suite of cloud services; this is increasingly important as AI diffuses across enterprise
- >90% of Reserved Instance customers consume at least two of our products and >75% use at least three
-These non-GPU services are increasingly contributing to CoreWeave's top and bottom-line, eclipsing $400 million of ARR in Q2. We expect this to be a major growth area across a broad array of customers in the coming quarters / years.

Every one of these workloads needs infrastructure built for AI from day one. That's what we built. And adoption rarely stops at the first contract. Customers put the compute to work, then come back for more. $CRWV

.@SemiAnalysis_'s ClusterMAX is the closest thing the industry has to an independent report card for GPU clouds. CoreWeave is the only provider rated Platinum (ie the highest tier) in all three consecutive evaluations. Proud of this team for continuing to raise the bar. $CRWV

investors.coreweave.com/news/news-deta…

New $CRWV customer. Love to see it.

The last few weeks have been a whirlwind of events, meetings, and calls. The thread running through every conversation is the same: momentum. The pace inside this business right now is unlike anything I’ve seen. Here’s what’s actually driving it:

We are still in the early days of AI. There is so much to build and so many incredible innovators building it. Demand continues to exceed supply as the agentic era ramps and the application layer matures. We believe we’re years away from closing that gap.

Pacing the frontier doesn't mean that AI adoption will slow - in fact it is likely to accelerate. As @GavinSBaker pointed out, implementing safeguards can itself require more compute.

We believe more than half of our workloads today are inference. Models aren't just being trained anymore, they're powering applications and agents being used at scale every day.

I believe the horizon is bright. Realizing that potential will take substantially more infrastructure. Our focus stays the same: building and operating that infrastructure reliably and securely, at scale.

$CRWV

When I talk with investors about AI infrastructure, GPUs are the starting point. Durability comes from making them useful: software, security, & expertise that help customers build and run products. That’s the business we’re building at CoreWeave. $CRWV

coreweave.com/blog/an-ai-clo…

More time for science. With Parallel Works, we’re giving DARPA’s NODES researchers access to a fully managed AI cloud so they can start training models and running biological simulations in days. Proud to see CoreWeave supporting this work. $CRWV

investors.coreweave.com/news/news-deta…

Well deserved. Grateful to continue building $CRWV with Mike and team.

Most complex R&D workloads are still constrained by standard cloud infrastructure. @RescaleInc is changing that by expanding with $CRWV as their AI platform. Their customers in aerospace, automotive, energy, and life sciences get compute that matches their workloads.

utm.io/uspUk

Enjoyed diving into $CRWV with @admaniexplains on @rundowndaily_. We covered a lot of ground but the key takeaways:

-enterprises building real AI applications need more than GPUs
-they need a full AI cloud platform: CPUs, memory, storage, high-performance networking, an incredibly robust management suite that enables GPU uptime
-all built to run training and inference on the same infrastructure

That’s what I believe we do better than anyone else, and it’s what true enterprise AI adoption requires. This and more in my chat with Zaid.

youtu.be/eoGzrS9cj6E?si…

Q2 revenue $2.6B, up 112%. Backlog $104B not including $25 billion in net new customer commitments in early Q3. 1.5GW of active power. 7 new AI platform capabilities. Executing, scaling and innovating, all while operating leverage accelerates. Built for pioneers by pioneers. $CRWV

utm.io/ur9Mo

Closed a $2.6B facility. For the first time in AI infrastructure, lenders are financing customer contracts that are shorter dated than the debt itself. That’s real confidence in the long-term value of NVIDIA GPUs running on CoreWeave. $CRWV

investors.coreweave.com/news/news-deta…

IMC is significantly scaling its use of @CoreWeave to power its quantitative research, moving from production since 2025 to a much larger commitment. Another financial firm going deeper with us, after Jane Street and Flow Traders. $CRWV

coreweave.com/news/imc-selec…

Storage is becoming a real constraint in AI capacity planning. Today we locked in multi-year priority access to @Solidigm enterprise SSD capacity. $CRWV

coreweave.com/news/coreweave…