The Economist's Impact 2nd AI Compute Summit - Smarter systems, leaner models: reducing compute costs without sacrificing quality

Announcement, slides.

At The Economist's Impact 2nd AI Compute Summit the conversation revolved about AI's most pressing issues: scarcity. Be it power, hardware, cost, speed, or know-how, the explosion of LLM's applications is vacuuming up resources on all fronts.

So how can companies get the most out of what they have? Is exponentially increasing spending necessary to keep up with the state of the art?

In my talk I explored the art of building smarter systems with leaner models: how to use efficiently the resources you have while getting cutting-edge performance out of them. The talk is very practical and highlights four levels for optimization, from context engineering down to your silicon strategy, and defines clearly the steps to take to go from a simple, inefficient prototype to a scalable product.