Early-stage venture capital is finally waking up from the software fever dream, aggressively recalibrating toward deep tech, physical infrastructure, and heavy hardware. According to a TechCrunch survey of early-stage investors evaluating the latest Y Combinator batch, funding has decisively pivoted away from generic software wrappers toward startups tackling tangible physical bottlenecks.
While participating investors noted that valuations are considerably more grounded than in previous speculative frenzies, the technical ambition of the standout companies remains unapologetically audacious. As one venture capitalist told TechCrunch, the pitches felt “like science fiction.”
Hardware and Energy Infrastructure
Severe power shortages and compute scaling limits have forced early-stage capital straight into complex industrial engineering. A prime example is Atomarine, which attempts to solve data center energy constraints by planning nuclear-powered facilities floating at sea, utilizing seawater for direct cooling. The company targets a gas-powered pilot launch by 2028 before attempting the transition to floating nuclear power ships by 2032.
Atomarine claims it has already secured over $4 billion in customer interest through letters of intent.
That staggering commercial interest instantly propelled Atomarine into the upper tier of the batch's valuations. Yet beneath the impressive letters of intent lies a stark reality: early-stage founders are now pitching roadmaps spanning multi-year timelines and billions in non-binding commitments, while their operational proof remains years away.
For executives navigating AI infrastructure, this shift is telling. The market is no longer impressed by another thin software layer burning venture cash on customer acquisition. Capital demands solutions to the physical realities of the compute bottleneck, even if those solutions require waiting a decade for a nuclear reactor to hit the water.