Richard Sutton the University of Alberta professor whose 1984 PhD thesis laid the groundwork for modern reinforcement learning has launched Oak Lab. The new Edmonton based outfit aims to build AI agents capable of human level reasoning while running on just 20 watts the power of a dim lightbulb. Backed by undislcosed funding the lab is hiring 25 researchers with a focus on energy efficient algorithms not bigger data centers. Sutton’s co founders include former DeepMind and Google Brain scientists who left to pursue what they call the next paradigm in AI.

This is a direct challenge to the orthodoxy that more compute equals better AI. Sutton’s work implies that the industry’s $100B annual capex splurge on GPUs and data centers may be a local maximum not a path to AGI. If Oak Lab succeeds even partially it could devalue Nvidia’s moat and force Big Tech to rethink their cloud AI strategies. More broadly it validates a growing school of thought that the next AI breakthrough will come from algorithmic efficiency not just raw power.

Winners include Oak Lab’s backers and Canada’s AI ecosystem which gets a marquee name. Losers could be Nvidia AMD and cloud providers if Sutton’s approach gains traction. Big Tech’s AI divisions at Google and Meta may also face pressure to diversify their R&D beyond scaling laws.

Watch for Oak Lab’s first peer reviewed papers expected within 18 months. If they demonstrate 10x efficiency gains on standard benchmarks expect a talent exodus from scale obsessed labs. Also track energy regulators who may start favoring efficient AI in policy.