What to expect during CoreWeave’s ‘Fully Connected’ event: Join theCUBE Sept. 30-Oct. 1
SiliconANGLE Mark Albertson
CoreWeave is kicking off its first user conference in San Francisco on Sept. 30-Oct. 1. The draw: new AI tools, Nvidia ties, and a backlog that’s already about $104 billion.
Based on reporting by SiliconANGLE, Mark Albertson — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
CoreWeave is using its first user conference to make a simple pitch: if you want AI infrastructure without building a giant pile of your own, this is the place to look. The company, which started in 2017 under another name and pivoted in 2019 from crypto mining to renting out surplus GPU power, has spent the last few years turning that idea into a cloud business for AI training, inference, and agent deployment.
Two recent launches tell the story. In May, CoreWeave rolled out a way for enterprises to deploy AI agents that can learn from real-world data and improve on their own. This month it said it will add a Physical AI Field Engineering service meant to connect industrial expertise with applied machine learning. Both moves point to the same goal: make AI less painful to build, cheaper to run, and faster to ship.
That cost argument is clearly part of the company’s sales pitch. John Furrier of theCUBE pointed to the growing talk around “asset light” AI, meaning companies don’t need to spend “billions and billions” to get intelligence working. CoreWeave’s own numbers are more concrete. The company says its autonomous improvement capabilities for AI agents could cut costs by more than 40% and speed training by about 1.4 times, without hurting quality.
CoreWeave’s pitch also rests on reliability and systems work, not just raw capacity. Chief Marketing Officer Jean English said customers want a partner that can handle training and inference at the pace they need, without latency or observability gaps. The company says that’s why it leans so hard on hands-on support and validation.
Nvidia is expected to loom large at the event, and not by accident. In early June, CoreWeave said it had completed what it described as the industry’s first bring-up and validation of Nvidia Vera Rubin NVL72 on CoreWeave Cloud, including liquid cooling, rack control, networking, and secure multi-tenant operations. And the demand picture is hard to ignore: in August, CoreWeave said its revenue backlog was about $104 billion as of June 30, before more than $25 billion in net new customer commitments added in early third quarter.
My take — AI-written commentary, not fact-checked reporting
CoreWeave is selling the boring truth of AI: most companies don’t want a model, they want the thing to work on Tuesday. That’s why the Nvidia validation and the “asset light” pitch matter more than the shiny agent talk. The industry keeps pretending infrastructure is plumbing; CoreWeave is happily charging rent for the pipes.
Read more about this at: SiliconANGLE