The conversation around AI’s future is shifting from model parameters to megawatts, and TechCrunch Disrupt 2026 is putting that reality center stage. The newly announced Smart Systems Stage lineup, running October 13-15 in San Francisco, makes one thing brutally clear: the most critical bottleneck for the next wave of AI isn’t better algorithms. It’s whether the lights can physically stay on.
The speaker roster reads like a who’s-who of the people actually trying to build a solution rather than just pitch one. Helion CEO David Kirtley and Commonwealth Fusion Systems Chief Science Officer Brandon Sorbom will share the stage to dissect what it actually takes to move commercial fusion from a lab experiment to a grid-scale power source. They’ll be joined by Jeff Lawson, the former Twilio founder who is now pouring his operational expertise into Inertia, one of the best-funded fusion startups around. His fireside chat promises a candid look at how scaling a SaaS giant shapes his approach to a hardware problem with no room for error.
But fusion is the long game. The more immediate crisis is the creaking infrastructure we already have. A panel featuring Drew Baglino of Heron Power and WeaveGrid’s Apoorv Bhargava will dig into the messy, unglamorous work of rewiring a grid that wasn’t built for the gigawatt-scale demands of modern data centers. That thread continues with a separate session bluntly titled “AI’s Power Problem,” where Bloom Energy’s Bill Thayer and Ambrosia Energy’s Sara Spangelo will discuss the frantic race to secure electricity before it smothers the very models it’s meant to support.
For founders and operators tired of hype cycles, this track cuts through the noise. It’s an acknowledgment that the next great infrastructure build-out won’t be in the cloud—it will be in copper, steel, and plasma physics. And with the current ticket pricing window closing soon, the event is positioning itself as the place where software’s future meets hardware’s unforgiving present.
