AI

Ep 94: Applied Compute CEO on the Limits of RL, the New AI Hyperscaler & Why Post-Training Wins Inference

Read full article at source ↗

Based on the episode description, not a full transcript. Source evidence retrieved 06/10/2026, 13:04:25 UTC.

Briefings

Yash Patil of Applied Compute argues post-training is the critical lever for AI ownership, distinguishing between optimal training and optimal serving. He identifies the largest inference workloads as the most valuable targets for this approach. The podcast covers the strategic shift from CPU to GPU-based AI hyperscalers, the challenges of reinforcement learning, and the economic implications of the Jevons paradox. Key figures include Yash Patil, Jacob Effron, and Dylan Patel, with a focus on the technical and economic realities of modern AI infrastructure.

Key points

• Yash Patil of Applied Compute identifies post-training as the most overlooked part of the AI stack, asserting it is the key to owning intelligence and distinguishing between optimal training and optimal serving.
• The podcast notes that the largest inference workloads are simultaneously the most valuable to train and the most valuable to serve, making them the primary targets for post-training optimization.

Context and limitations

Evidence is limited to a single podcast episode discussing AI strategy, technical architecture, and economic factors. Specific company financials, exact model performance metrics, or future predictions are not provided.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 06/10/2026, 13:18:46 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeihpwoljmdcpfsv65w35n5cqv6qq3kfhpacu2ex7vcckvlmutfg74y