The biggest AI models are no longer the only option

10 Aug 2026 · 1 min

Watch on YouTube

0:46

In this solo talk, Dan Novaes challenges the long‑standing belief that only massive, data‑center‑scale AI models can deliver high performance. He outlines a growing shift toward smaller, more efficient models that can run directly on consumer devices such as laptops and smartphones. By moving inference to the edge, these models offer faster response times, lower operational costs, and stronger privacy because data can stay on the device. Novaes argues that this trend will dramatically broaden AI access, especially for users in regions with limited bandwidth or expensive cloud services. While the largest models will retain certain advantages, the future may be defined by “smart” models that achieve more with less computational resources.

Chapters

For years, everyone believed AI had to be massive. Huge data centers, thousands of powerful chips, billions of dollars— that was the only way. Now, that's starting to change. AI models are getting smaller, more efficient, and surprisingly capable. Some can now run directly on a laptop or even a smartphone without sending every request to the cloud. That means faster responses, lower costs, and in many cases, better privacy because your data doesn't always have to leave your device. This shift could make AI available to billions more people, especially in places with slower internet or limited access to expensive cloud services. The biggest AI models will still have an advantage, but they're no longer the only option. The future of AI may not belong solely to the largest models. It may belong to the smartest ones, the ones that can do more with less.