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

October 6, 2026

AI Summary

5 min read

In the AI world, the narrative around "owning your intelligence" has shifted from a defensive posture—fear of labs pulling APIs or training on your data—to a more pragmatic one about flexibility and control. Josh from Applied Compute, a company that helps businesses post-train open-weight models, argues that the real advantage isn't about avoiding malicious actors but about gaining the ability to decide where models run, how they are optimized for cost or latency, and how they are tuned to specific domains. This conversation digs into the mechanics of when and why post-training makes sense, the limits of current reinforcement learning (RL), and why the biggest gains often come not from chasing frontier capabilities but from pushing out the price-performance Pareto curve.

The Hill-Climbing Machine: RL's Promise and Its Bottleneck

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

Never miss an episode of Unsupervised Learning with Jacob Effron

Get every new episode summarized in your inbox — free, ~5 minutes to read.

No spam. Unsubscribe anytime.

What you'll learn

1 (00:02) **Current State of RL and Own Your Intelligence** - The conversation opens with the guest, Josh from Applied Compute, defining RL as a "hill climbing machine" and arguing the hardest part is defining the hill (evals). He then frames the "own your intelligence" narrative.
2 (04:49) **The Multi-Model Future and the Pareto Curve** - Josh discusses the shift from a single "best model" narrative to a multi-model future where cost-performance trade-offs matter, using the OpenAI-Base 10 partnership as a key signal.
3 (07:23) **Post-Training for Capabilities vs. Cost/Speed** - The conversation focuses on when post-training can beat frontier models on raw capabilities, not just on cost and speed.
4 (09:23) **The Future of Proprietary Data and Online RL** - Josh predicts how the nature of training will evolve from offline RL to online RL, using production inference to continuously improve models.
5 (12:28) **Data Efficiency and the Road to Continual Learning** - Josh addresses the challenge of needing large amounts of data for RL and explains the path toward more data-efficient methods.
6 (14:50) **When to Train vs. When to Configure** - The discussion explores when it makes sense to invest in training a model versus using prompt engineering, context, or other "harness" optimizations.
7 (16:10) **Real-World Example and TCO of Post-Training** - Josh provides a concrete example of using user feedback to bias a model and discusses the total cost of ownership (TCO) for post-training.

+ Full timestamped outline available in the app

Show Notes

Yash Patil, a former OpenAI researcher who worked on Codex and now runs Applied Compute, argues that post-training is the most overlooked part of the AI stack and the key to owning your intelligence.

He explains when companies should optimize their harness and context first, and when it's worth updating the weights: the largest inference workloads, plus tasks that change over time or depend on judgment. That's the heart of why he believes post-training wins inference. The biggest workloads are both the most valuable to train and the most valuable to serve, and how a model is trained shapes how it should be served. Getting production serving running is the easier half; echoing Dylan Patel, he says GPUs plus vLLM plus OpenRouter "kind of" gets you there.

Yash is candid about the limits of RL. It's a hill-climbing machine whose hardest part is defining the hill, which makes a company's private evals as worth protecting as its employees. It generalizes far less than pre-training. And continual learning remains blocked by the unsolved problem of extremely data-efficient training from sparse rewards.

He lays out his bet on a new AI hyperscaler, built on GPUs the way AWS, GCP, and Azure were built on CPUs. He pictures it as an inverted pyramid with training at the bottom, where few can compete, and inference, routing, and harness stacked above. The conversation also covers whether every firm needs its own model, why Jevons paradox has made cost matter more than he expected, how models reward hack "like water," and why the US needs American open-weight models even though "blanket bans are not the answer."

(0:00) Intro
(1:42) The Case for Owning Your Intelligence
(4:35) OpenAI x Baseten and the Multi-Model Future
(7:19) Can Post-Training Beat Frontier Models?
(9:03) What Data Stays Out of Distribution
(11:51) Data Efficiency and Continual Learning
(14:31) When Updating Model Weights Makes Sense
(16:01) Online Training in Practice
(18:16) Lowering the Cost of Post-Training
(19:48) Does Every Company Need Its Own Model?
(24:06) RL in Non-Verifiable Domains
(26:59) How Applied Compute Picks Customers
(28:46) Building an Inference Business
(34:43) The Case for a New AI Hyperscaler
(38:21) Building RL Environments
(41:33) Quickfire

Jacob is an AI investor at Redpoint Ventures. He's led Redpoint's investments in companies like Abridge, Physical Intelligence & Legora. Follow Jacob on Twitter (@jacobeffron).

On Unsupervised Learning we probe the sharpest minds in AI in search for the truth about what's real today, what will be real in the future and what it all means for businesses and the world. If you're a builder, researcher or investor navigating the AI world, this podcast will help you deconstruct and understand the most important breakthroughs and see a clearer picture of reality. Subscribe to this show to stay up to date on our latest episodes.