Our Take
Barrett is right that venture capital abandoned the physical stack for a decade, but his portfolio's wins (Manifold Bio, PsiQuantum) prove the thesis, not the moment—the real risk is that capital flooding in now will chase the same inefficiencies it just corrected.
Why it matters
As AI demand for chips and power outstrips supply, venture is repricing hardware and deep tech after years of neglect. Practitioners need to understand where capital is moving and which sectors are now crowded versus still underexplored.
Do this week
Infrastructure teams: audit your energy and semiconductor roadmaps for the next 24 months against Playground's stack (nuclear to in-chip power delivery)—if you're not planning for 100x efficiency gains, your cost model is already stale.
Playground Global Closes $475M Hardware-First Fund
Playground Global, the deep-tech venture firm founded in 2015, closed a $475 million fund focused on seed and Series A investments in semiconductors, quantum computing, robotics, energy infrastructure, and materials science (company-reported). The firm added Pat Gelsinger, former CEO of Intel and VMware, as a general partner specializing in semiconductors, joining co-founder Peter Barrett and other partners Jory Bell and Bruce Leak.
The fund reflects a decade-long thesis that breakthroughs in hardware and physics, not just software, would define the next generation of valuable companies. Barrett argues the venture industry is now catching up to what Playground identified early: as AI demand for compute explodes, the energy and semiconductor layers that software depends on have become the actual constraint.
Capital Is Finally Repricing the Physical Layer
For much of the 2010s, Silicon Valley chased software and apps while venture capital starved hardware companies of attention. Playground's portfolio companies illustrate what that capital drought cost. Manifold Bio, which uses in vivo testing to generate pharmaceutical datasets, closed a deal with Roche valued at up to $2 billion last year. PsiQuantum, a quantum computing startup that began with only three employees in Playground's Palo Alto lab, is now building superconducting quantum machines described as 10,000 times the state of the art when the firm first backed it.
The shift is structural, not cyclical. AI is consuming electricity at a rate that forces every operator—from hyperscalers to Fortune 50 companies—to rethink power delivery from reactors down to in-chip voltage conversion. Playground's stack spans nuclear reactors to memory systems to superconducting logic (Snowcap), a company whose devices switch five orders of magnitude more efficiently than transistors, though at cryogenic temperatures.
Barrett is candid about the risk: much of the capital now chasing hardware will be deployed inefficiently. Venture practice in deep tech requires domain expertise that cannot be spun up overnight. He also dismisses a category of projects that have good physics but bad economics—data centers in space, small modular reactors, fusion—as sensible to avoid.
Where the Capital Is and Isn't Going
Playground's fund composition signals which sectors are now crowded and which still underexplored. The firm is openly investing across the compute stack: chip architectures, power delivery, materials science, and quantum algorithms. It has also backed robotics (Agility) and aerospace manufacturing, areas that benefit from the same shift toward physicality.
Barrett's dismissal of space-based data centers and his emphasis on 100x to 1,000x efficiency gains in general-purpose compute point to a narrower opportunity: not all hardware gets funded equally. Companies solving efficiency—whether through superconducting logic, novel chip architectures, or power infrastructure—are more likely to see capital than moonshot projects with unresolved economics.
The firm has also emphasized portfolio construction that mixes tactical and strategic bets. Some portfolio companies (like the Manifold Bio example) reach exit within a handful of years. Others, like PsiQuantum or Snowcap, may take a decade to maturity. This portfolio diversification is not novel, but it is a deliberate hedge against the assumption that all deep tech requires 10+ year horizons.