로보틱스 및 피지컬 AI 예측: 2026년의 6가지 투자 전망

The robotics industry is entering a 'GPT-2.5 moment' where Physical AI foundation models begin to demonstrate scaling laws, shifting the focus from simple control algorithms to complex physical understanding. As data collection costs are projected to exceed $3 billion, 'World Models' that learn physics from internet-scale video are emerging as the primary solution to overcome the structural scarcity of robot-specific data. This high-capital requirement, combined with extreme talent concentration in top-tier institutions, is rapidly accelerating a winner-take-all dynamic that favors established players with massive compute and data resources.