Gradient Dissent: Conversations on AI Inside the $41B AI Cloud Challenging Big Tech | CoreWeave SVP
37 snips
Jan 6, 2026 
Corey Sanders, SVP of Product at CoreWeave and former Azure executive, discusses the revolutionary 'Neo Cloud' model essential for modern AI workloads. He explains how general-purpose clouds falter when faced with the unique demands of AI, emphasizing the need for specialized solutions. Corey highlights CoreWeave's innovative technology for optimizing GPU throughput and the importance of customer engagement in product development. He also shares insights on infrastructure design, liquid cooling, and the significance of tailored cloud strategies.
AI Snips
Chapters
Books
Transcript
Episode notes
Low-Budget Videos To Tell Local Stories
- Corey filmed early cloud culture videos on his cell phone in local data center cafeterias to tell customer stories.
- That low-budget, customer-focused approach increased authenticity and engagement.
AI Needs GPU-Centric Infrastructure
- AI workloads require feeding GPUs with massive data throughput to avoid idle expensive accelerators.
- Designing infrastructure specifically for AI enables assumptions public clouds struggle to make.
Specialization Enables Unique Data Center Design
- Specialization permits design choices like liquid cooling and tailored object storage that public clouds can't assume everywhere.
- Those design choices materially improve efficiency and enable newer, denser GPU deployments.
