Google Developing "Frozen v2" Next-Gen AI Chip with Up to 10x Efficiency Gain
According to an exclusive July 20 report from The Information, Google is developing a next-generation server AI chip codenamed "Frozen v2" that hardens core Gemini model computing architectures directly into hardware, drastically cutting inference costs by reducing runtime decisions and data movement. Sources reveal the chip will deliver 6 to 10 times better energy efficiency than Google's current most powerful custom AI chips, with deployment targeted as early as 2028. Following the report, Alphabet shares rose as much as 3.7% intraday. Industry observers note this signals AI competition among tech giants has expanded from model algorithms to full-stack infrastructure rivalry.
#GoogleAI #Frozenv2 #AIChip #Gemini #ComputeInfrastructure
new
AMD Unveils Helios Rack-Scale AI System, Microsoft Azure First Major Customer
Google Developing "Frozen v2" Next-Gen AI Chip with Up to 10x Efficiency Gain
Honor Launches World's First Robot Phone, Ushering in Embodied Interaction Era
Tencent Upgrades Embodied AI Stack, Hunyuan Hy3 Tops Global Call Volume
Moonshot AI Unveils Kimi K3: World's Largest Open-Source Model at 2.8T Parameters
Apple Intelligence Completes China Regulatory Filing, Alibaba Qwen Integrated Across Apple Ecosystem
China's Self-Developed AI Chip Achieves Architectural Breakthrough – 520 TFlops at 14nm
US Media Outlets Seek Severe Sanctions Against OpenAI in Copyright Lawsuit
Google Cloud CEO: Constructive EU Talks Underway to Advance AI Content Watermarking
Nubia Unveils World's First Mass-Produced AI Agent Smartphone, Debuting at WAIC 2026