VCs pour $47.4B into physical AI in H1 2026, nearly 4× prior half

The surge in physical AI funding reflects a fundamental shift in how venture capitalists view the next frontier of AI deployment. After years of pouring capital into large language models and software-only AI platforms, institutional investors are now placing billion-dollar bets on companies that embed AI into hardware, sensors, and real-world operational systems. The timing is not accidental: the economics have finally aligned. Hardware costs have declined substantially—even consumer mobile phones now carry LIDAR scanners—while compute and foundation-model capabilities have become more accessible and reusable. Training data is more plentiful, and sensor costs continue to fall, making it feasible for smaller teams to build and scale physical AI companies with less capital than was previously required.

What distinguishes this wave from prior robotics or automation booms is the business model. Rather than selling expensive one-off machines, companies are increasingly bundling hardware into recurring or mixed-revenue models and adopting outcome- or usage-based pricing. This transforms the hardware itself into a distribution mechanism for software and data services, creating what investors call a "data intelligence flywheel." Vertically integrated companies that own multiple layers of the stack—hardware, sensors, software, and data—are seen as likely to capture the most value. Industries like manufacturing, supply chain, utilities, agriculture, and defense have become attractive precisely because they are traditionally capital-intensive and analog, making AI-driven efficiency gains (predictive maintenance, autonomous operations, risk management) measurable and commercially compelling. The convergence of cheaper tech, experienced talent migrating from software, rising market demand, and supportive policy has created conditions that venture firms believe will sustain investment at record levels for years.