英伟达CEO黄仁勋撰文：还需要投资数万亿美元建设AI基础设施

Core Insights - AI is recognized as a powerful force shaping the modern world, essential as infrastructure like electricity and the internet, rather than just clever applications or models [2][20] - The AI industry is still in its early stages, having seen investments of hundreds of billions, but requires trillions more for foundational infrastructure development [1][31] - AI's architecture is defined as a "five-layer cake," which includes energy, chips, infrastructure, models, and applications, each layer reinforcing the others [11][30] Group 1: AI Infrastructure - The foundational layer of AI is energy, which is crucial for real-time intelligence generation, making it a bottleneck for the system's intelligence production capacity [7][25] - Above the energy layer is the chip layer, which focuses on efficiently converting energy into computational power, essential for AI workloads that require high parallel processing capabilities [8][26] - The infrastructure layer encompasses the physical systems needed for AI factories, including land, power supply, cooling systems, and network communication [9][27] Group 2: AI Models and Applications - The model layer is capable of understanding various types of information, with significant advancements occurring in fields like protein AI, chemical AI, and robotics [10][28] - The application layer generates economic value through various implementations, such as drug development platforms and autonomous vehicles, showcasing the diverse outcomes of the same architectural framework [10][29] Group 3: Economic Impact and Workforce - The ongoing construction of AI factories represents one of the largest infrastructure projects in human history, requiring a vast workforce across various technical roles [12][32] - AI is enhancing productivity across knowledge economies, as seen in radiology, where AI assists in image interpretation while increasing the demand for radiologists [12][32] - The past year has seen significant improvements in AI model performance, leading to real economic value creation in sectors like drug development and logistics [14][34] Group 4: Future Directions - The current phase of AI development is characterized by a need for continued investment and workforce training, with many opportunities yet to be explored [17][36] - The direction of AI as a foundational infrastructure will depend on choices made today regarding its construction, deployment, and responsible use [18][38]