# Inside the $41B AI Cloud Challenging Big Tech | CoreWeave SVP

Page: https://stenobird.com/podcast/gradient-dissent/inside-the-41b-ai-cloud-challenging-big-tech-coreweave-svp
Text version: https://stenobird.com/podcast/gradient-dissent/inside-the-41b-ai-cloud-challenging-big-tech-coreweave-svp.md
Podcast: [Gradient Dissent: Conversations on AI](https://stenobird.com/podcast/gradient-dissent)
Published: 2026-01-06T13:00:00+00:00
Episode link: https://wandb.ai/site/resources/podcast
Audio file: https://episodes.captivate.fm/episode/167db7d2-6e2f-406e-b26b-19e329f838c8.mp3
Processing state: processed
JSON: https://stenobird.com/v1/public/podcasts/gradient-dissent/episodes/inside-the-41b-ai-cloud-challenging-big-tech-coreweave-svp
Duration seconds: 3199

## Resource

General-purpose clouds are struggling to meet the specialized demands of large-scale AI training. CoreWeave's approach focuses on maximizing GPU throughput and observability to solve the unique bottlenecks of modern workloads.

## Highlights
- Main idea: The shift toward AI workloads requires a 'Neo Cloud' model that prioritates GPU throughput over API consistency
- Practical takeaway: Maximizing 'goodput' requires specialized hardware considerations like liquid cooling and optimized data paths
- Failure mode: General-purpose clouds fail when they cannot make the specific hardware assumptions necessary for massive AI training jobs
- Main idea: Customer success in AI infrastructure depends on deep observability and automated orchestration like Slurm on K8s
- Practical takeaway: High-quality product feedback is best gathered by presenting functional, high-fidelity prototypes rather than abstract ideas

## Topics

AI Infrastructure, GPU Orchestration, Cloud Computing, Machine Learning Operations, CoreWeave, Data Throughput, Cloud Strategy, Kubernetes

## Chapters
- 1:00 — Introduction: Lukas Biewald introduces Corey Sanders, SVP of Product at CoreWeave, discussing his transition from Azure to the specialized AI cloud space.
- 9:00 — The Evolution of AI Workloads: A discussion on how the AI revolution has created a new tier of business-critical workloads that mirror the previous analytics wave.
- 13:05 — CoreWeave's Technological Innovations: Exploring how specialized assumptions regarding hardware, such as liquid cooling and optimized data movement, provide a competitive edge.
- 25:05 — Customer Engagement and Future Prospects: How CoreWeave uses deep observability and technical customer enablement to drive product adoption and user satisfaction.
- 29:05 — Comparing Cloud Approaches: Contrasting the 'Neo Cloud' model with traditional public cloud strategies regarding customer service and infrastructure flexibility.
- 37:00 — Balancing Executive Roles and Hands-On Projects: Corey discusses the importance of staying close to the technical implementation and the challenges of GPU availability.
- 41:20 — Product Development and Customer Feedback: The value of using high-fidelity mockups and functional prototypes to elicit actionable feedback from engineers.

## Actions

- request_transcript: `POST https://stenobird.com/v1/public/podcasts/gradient-dissent/episodes/inside-the-41b-ai-cloud-challenging-big-tech-coreweave-svp/transcription-requests` — Idempotently request low-priority transcript generation for this episode.
- read_markdown: `GET https://stenobird.com/podcast/gradient-dissent/inside-the-41b-ai-cloud-challenging-big-tech-coreweave-svp.md` — Read the agent-friendly Markdown representation of this episode resource.

A page view does not enqueue transcription. Agents should invoke `request_transcript` explicitly when they need this episode processed.

## Transcript

Full transcripts are not published on public pages unless there is a clear rights basis.