Analysis
SE Ventures, the corporate venture arm of industrial giant Schneider Electric, published a new investment thesis this week arguing that the AI buildout's physical constraints -- grid power, thermal management, interconnection capacity -- are creating a durable industrial investment cycle that most venture capital remains underweight relative to its scale.
The firm's argument, laid out by a Schneider Ventures partner, is that the software and chip layers of the AI stack get most of the venture attention and headlines, while the physical infrastructure layer -- the power electronics, cooling systems and grid hardware that actually let a data center run at gigawatt scale -- is comparatively starved of specialized capital despite representing an enormous and rapidly growing spending category.
Schneider's dual role as both an industrial conglomerate selling directly into hyperscale data center buildouts and an active venture investor gives the firm an unusually direct line of sight into which startups are genuinely solving hard physical problems -- like grid interconnection delays and liquid cooling at scale -- versus which are packaging existing approaches with an AI narrative attached.
The thesis lands amid a broader wave of corporate venture commentary this year arguing that AI's most durable investment opportunities increasingly sit in unglamorous, capital-intensive physical infrastructure categories -- power, robotics, energy storage -- that traditional software-focused venture firms are structurally less equipped to evaluate and fund at the check sizes these categories require.
What to watch: which specific startups SE Ventures backs in the coming quarters as a signal of where it sees the strongest physical-infrastructure opportunities, whether other industrial conglomerates follow with similar dedicated AI-infrastructure venture theses, and whether pure-software VCs begin partnering more actively with industrial corporate investors to access this deal flow.
