Key Takeaways

  • Venture capital has concentrated heavily, shifting from tracking a hundred promising early-stage bets to funding the top 10 mega-cap leaders ahead of public market listings.
  • Hardware companies spend five to eight years storing potential energy through deep research and development before that work suddenly converts into kinetic market energy.
  • Sequoia's core early-stage filter asks one question: would you allocate 100% of your career equity to work for this founder?
  • Winning in physical AI requires synchronizing startup development cycles directly with market conversion timelines.
  • Evaluating physical bottlenecks requires David Cahn's 3 Pillars of AI Data Center Infrastructure.

The 3 Pillars of AI Data Center Infrastructure

  • 1. Server (Compute): Chips and GPU compute hardware that power the models (the primary focus of current market capex).
  • 2. Steel (Industrial Infrastructure): The physical and industrial manufacturing components needed to assemble physical data center facilities, such as gas turbines and specialized hardware components.
  • 3. Power (Energy and Grid): Grid-scale energy capacity, grid interconnects, long-duration battery storage (such as iron-air batteries), and nuclear power needed to sustain data center electricity requirements 2 to 4 years out.

When This Works (and When It Doesn't)

Cahn uses this checklist to look past software wrappers and evaluate where capital actually creates defensible moats across the physical supply chain. Most venture dollars still chase compute chips and foundation models. The real bottleneck is shifting downstream into heavy industrial assembly and raw megawatts. If a startup solves grid interconnects, gas turbine supply, or multi-day battery storage, it bypasses the crowded software layer entirely.

This framework breaks down if you treat energy timelines like software cycles. You cannot sprint your way through regulatory approvals for nuclear plants or utility interconnect queues. As Cahn notes, “In hardware there's potential energy. And these hardware companies spend five, six, seven, eight years building this potential energy. And it's because they're doing real very difficult R&D work and at some point that potential energy converts into kinetic energy.” If your company builds deep hardware, your financing must match that multi-year lag before the market converts.

Beyond hardware supply chains, the ultimate bottleneck remains talent gravity. As Cahn puts it: “The framework we often use is would you go work for this person? Because I think everything is sort of subsumed in that. If other people have to be willing to go work for this person, the decision of whether to go work for a person is effectively putting 100% of your equity in this one company.” If the founder cannot convince tier-one engineers to quit their jobs and risk their careers, no amount of capital can fix the gap.

What to Do With This

Run the career equity audit on your own company this week. Pull up the profiles of the last three senior engineers or executives you failed to close.

Ask yourself honestly: did they pass because of cash compensation, or because they did not trust that betting 100% of their career equity on you would pay off? If it was the latter, your pitch is focusing too much on product features and not enough on the market inflection window.

Next, evaluate your product against the infrastructure pillars. If you rely entirely on third-party compute APIs without a defensible angle on cost, power access, or physical hardware efficiency, map out how a sudden compute crunch or electricity cost spike affects your margins over the next 24 months.