[Qiming Venture Partners Releases '2026 AI Top Ten Outlook' Covering Foundation Models, Embodied Intelligence, AI Infrastructure, and AI Applications] DeepFlow Tech News, July 19 — At the WAIC 'Qiming Venture Partners · Entrepreneurship and Investment Forum,' Qiming Venture Partners released the '2026 Qiming Venture Partners AI Top Ten Outlook,' which covers four areas: foundation models, embodied intelligence, AI infrastructure, and AI applications. In terms of foundation models, Qiming Venture Partners predicts that in the next 12 to 24 months, top-tier models will internalize most 'plug-in' capabilities, and multimodal models will further evolve toward interactive world modeling. Regarding embodied intelligence, the effective data scale of leading robotics companies is expected to increase from '10,000-hour level' in 2025 to '1 million-hour level' by 2027, while dexterous hand and tactile sensing technologies will mature rapidly and continue to reduce costs. For AI infrastructure, it is noted that the focus of AI computing power demand is shifting toward inference. Over the next two years, AI infrastructure may continue to face structural shortages, and safe and trustworthy AI will upgrade from being optional to mandatory. As for AI applications, in the next 12 to 24 months, AI application business models will accelerate the shift toward pricing based on outcomes and value. Commercialization will increasingly focus on vertical scenarios and high-paying users, while AI-Native organizations will transition from concept to empirical validation.