Pat Gelsinger Says the Next Leap in Computing Will Be Heterogeneous — and Investors Should Pay Attention

From CEO to investor

It has been reported that Pat Gelsinger, the one-time face of Intel’s resurgence efforts, is now working from Playground Global, trading product roadmaps for startup decks. At 65 he told Ian Cutress he wants “to do things that matter” — not exactly a retirement plan, more a second act with purpose. He’s no longer running Intel, but the themes are familiar: manufacturing, architectures, scaling. And yes, Moore’s Law still threads through his thinking.

The “trinity of computing” and the road ahead

In a wide-ranging interview on More Than Moore, Gelsinger argued that the future will be fundamentally heterogeneous — a “trinity” of classical, AI, and quantum systems — rather than another incremental shrink of the same CPU. He warned that inference efficiency needs orders-of-magnitude gains and flagged hotspots for disruption: AI accelerators, dataflow machines, optical links, resilient networking, and novel precision and resilience techniques. Agentic and scientific AI workloads? He thinks we’re still early. Strong words from someone who’s seen every major shift in compute — and that’s worth pausing over.

What he’s looking for — and why it matters

Shifting from CEO to evaluator changes the rubric. Gelsinger says Playground is hunting hard-physics bets: teams with technical depth, clear proof, and market fit — leaders who can scale. It’s not just VC boilerplate; he wants impact and people he enjoys working with. The industry should listen: if a veteran who once pushed “five nodes in four years” is betting on heterogeneity and dataflow, that’s a directional signal. Will the rest of the ecosystem follow? That’s the question now, as startups and incumbents race to make inference cheaper, systems more diverse, and the next computing paradigm practical.

Sources: morethanmoore.substack.com, Hacker News