Nvidia(US:NVDA) 虎嗅APP·2026-01-05 00:11Core Viewpoint - The article discusses the evolving landscape of AI hardware and embodied intelligence, highlighting the rapid commercialization and investment trends in these sectors, while also addressing the challenges and competitive dynamics faced by startups and established players in the market [4][6][30]. AI Hardware Industry - The concept of "AI hardware" has rapidly converged to include consumer electronics products that leverage model inference capabilities, such as AI glasses and AI toys, moving away from more speculative product ideas [4][5]. - The domestic AI hardware industry benefits from a mature supply chain, allowing entrepreneurs to effectively "recreate" consumer electronics, which has led to a unified approach among industry players [5][6]. - Despite the enthusiasm in the capital market for AI hardware, there is a notable absence of standout products that can validate the industry's direction [6][10]. - The AI glasses segment has seen significant growth, with shipments increasing from 1.3 million units in 2024 to 2.9 million units in 2025, reflecting over 120% year-on-year growth [16][17]. Embodied Intelligence Sector - The embodied intelligence sector has experienced a surge in valuations, with many companies reaching valuations in the tens to hundreds of billions, driven by a robust domestic supply chain and strong market demand [8][24]. - However, there is a disconnect between the high valuations and actual order disclosures, with many companies opting not to reveal specific order quantities or amounts [25]. - The primary application of humanoid robots has been in research settings rather than commercial production, which raises concerns about the sustainability of this business model [26][27]. - The emergence of robot rental services is a notable trend, with the market expected to grow from 1 billion to 10 billion in the coming year, indicating a shift in how humanoid robots are utilized [27]. AI Chip Industry - The domestic AI chip industry is experiencing a significant breakthrough, with the penetration rate of domestic chips rising from 8% to 23% in the past year, driven by government policies and the establishment of over 250 intelligent computing centers [32][33]. - Despite the growth, there are concerns about the global shift away from GPU-centric architectures towards more efficient alternatives like TPU and reconfigurable architectures, which may challenge the current dominance of domestic AI chips [34][36]. - The year 2026 is anticipated to be a pivotal moment for the domestic AI chip industry, as companies will need to compete on energy efficiency and performance against international brands [37].