半导体行业观察·2025-11-19 01:35Core Insights - GlobalFoundries has acquired Singapore-based silicon photonics manufacturer AMF, becoming a leading player in the silicon photonics sector [2] - Silicon photonics technology uses light instead of electrical pulses for data transmission, which is crucial for communication within chips, between components, and across servers [2] - NVIDIA plans to implement silicon photonics technology in its next-generation AI servers to reduce power consumption while increasing data transfer speeds, which is essential for managing millions of GPUs in data centers [2][3] - AMD is also investing heavily in this technology, having established a dedicated R&D center in Taiwan with an investment of nearly $300 million and previously acquiring Enosemi to enhance its competitive edge against NVIDIA [2][3] Industry Trends - The demand for silicon photonics technology is driven by the rapid expansion of AI data centers, which are experiencing significant increases in power consumption and placing pressure on electrical grids [3] - Tim Breen, CEO of GlobalFoundries, emphasized the importance of high-speed, precise, and energy-efficient data transmission for AI data centers and advanced telecommunications networks [3] - Silicon photonics is also a key technology for quantum computing, allowing for systems that do not require low-temperature cooling, making quantum computers more practical and cost-effective [3] Competitive Landscape - Major companies like AMD, NVIDIA, and GlobalFoundries are investing in silicon photonics, but several startups such as Ayar Labs, Celestial AI, and Lightmatter are also developing their own photonic technologies, indicating a competitive landscape [3] - While consumer-grade CPUs, GPUs, and motherboards are not expected to adopt this technology in the short term, it is already being utilized in large servers capable of processing terabytes of data per second [4] - Continued investment in AI data centers is anticipated to drive further research and development in this promising technology [4]