Hyperscalers Pay for Speed Over Efficiency in the Data Center Race
Leo Polovets explains why AI data center bottlenecks reward speed over efficiency, creating openings for startups in power, EPC design, and custom ASICs.
Why the Next Trillion-Dollar Startup Won’t Be a Software Company
With David Weisburd, Leo Polovets · Sunday, October 18, 2026
Leo Polovets, General Partner at Humba Ventures and co-founder of Susa Ventures, discusses why artificial intelligence is commoditizing software and shifting venture scale opportunities toward deep tech and physical hardware. Polovets explains how hardware companies scale differently than SaaS, addresses misconceptions about capital intensity, and breaks down the bottlenecks in AI data centers, energy, and semiconductors.
Key takeaways
- Hyperscalers face physical ceilings across energy grids, permitting, and engineering, forcing them to value fast deployment over efficiency. Read more →
- The traditional SaaS founding pair, splitting commercial duties and software engineering, breaks down in physical hardware and deep tech ventures. Read more →
- Software revenue ramps smoothly, but physical hardware runs near zero for two to five years before spiking straight into tens of millions of dollars upon deployment. Read more →
- Leo Polovets and Humba Ventures avoid early-stage humanoid robotics rounds despite predicting the category will produce a trillion-dollar winner. Read more →
- Leo Polovets points to the classic pottery class experiment to explain deal calibration: volume drives pattern recognition, while limiting dealflow to one curated pitch a week leaves investors poorly calibrated. Read more →
5 articles from this episode
Deep Tech Needs a Different Founder Archetype
Leo Polovets explains why physical tech kills the two-founder SaaS model and demands multi-disciplinary fluency across engineering and procurement.
Leo Polovets on Evaluating Pre-Revenue Deep Tech Startups
Leo Polovets explains how investors evaluate early-stage deep tech startups when revenue sits at zero for years before scaling vertically.
Leo Polovets on Why Humanoid Robotics Traps Early Venture Capital
Leo Polovets explains why Humba Ventures avoids humanoid robotics, warns against component suppliers, and bets on specialized industrial arms.
Leo Polovets: Passing on Price Was Susa's Costliest Mistake
Leo Polovets of Humba Ventures on why passing early-stage deals on price backfires, how meeting volume drives calibration, and why DPI trumps markups.
More How I Invest episodes
- Nelson Mills on Building a $100 Billion Company, the China Threat, and the Future of WarOct 18 · 4 articles
- John Borthwick on AI Agents, the Death of Apps, and the Future of SoftwareOct 18 · 6 articles
- Will Robbins on Elon Musk, AI & the Future of Silicon ValleyOct 11 · 6 articles
- Why This 22-Year-Old Founder Chose Defense Over Silicon ValleyOct 11 · 6 articles