Eight National Departments Jointly Issue Document to Strengthen AI Computing Power Supply: The Ministry of Industry and Information Technology and seven other departments issued the "Implementation Opinions on the 'AI + Manufacturing' Special Action." It explicitly calls for supporting breakthroughs in key core technologies such as high-end training chips and edge-side inference chips, and orderly advancing the layout of high-level intelligent computing facilities.
Domestic Computing Power Ecosystem Accelerates Path to Self-Reliance: Industry reports indicate that against the backdrop of intensifying global AI computing power competition, the domestic computing power industry is accelerating its autonomous development along the path of device, edge, and cloud. Related supply chain companies are drawing attention.
AI Data Center Super Cycle Seen as Key Investment Theme: Huatai Securities analysis suggests that the super cycle of hardware demand for computing, storage, and networking driven by AI data center construction remains the most important theme in the tech sector for 2026.
NVIDIA Unveils Next-Gen AI Platform 'Rubin': At CES 2026, NVIDIA officially launched its next-generation AI platform named Rubin. By combining new Vera CPUs and Rubin GPUs, among other components, the platform claims to reduce inference costs for some AI models by up to 90%, aiming to make enterprise-grade AI applications more economically viable.
AMD Announces Large-Scale AI Server Rack Solution: AMD plans to launch a server rack solution named Helios in the second half of 2026. This rack can accommodate up to 72 of its MI450 series GPUs. Oracle and OpenAI have become its customers, with Oracle planning to deploy tens of thousands of AMD chips starting Q3 2026.
Broadcom Benefits from Google's External TPU Sales: As a primary partner for Google's TPU (Tensor Processing Unit), Broadcom will benefit from Google's new strategy of selling TPU chips to third parties. Reports indicate that AI startup Anthropic has placed a $21 billion TPU order with Google.
Intel Introduces New GPU Targeting Inference Market: Intel plans to launch a new generation of data center GPUs named Crescent Lake. This product is primarily designed for the AI inference market, optimized for air-cooled servers, and positioned at a more economical price point.
Data Center Industry Undergoes Fundamental Transformation: Reports note that data centers in 2026 are evolving from "server rooms" into "computing power entities" that evolve in sync with AI chips. To meet AI computing power demands, single-rack power is rapidly increasing to 100-200 kW, and liquid cooling technology is transitioning from a 'cutting-edge tech' to the default configuration for new data centers.
AI Computing Demand "Crowds Out" Consumer Hardware Market: Market analysis points out that in building trillion-dollar-scale AI infrastructure, tech giants are essentially monopolizing production capacity for high-end chips and HBM (High Bandwidth Memory). This is squeezing supply and putting upward price pressure on consumer-grade graphics cards, memory, and other hardware, potentially impacting the entire consumer electronics industry.
Emerging Cloud Providers Shift Focus to AI Inference and Network Sovereignty Services: Facing competition from mainstream cloud providers, emerging cloud service providers (Neoclouds) specializing in GPU computing are adjusting their strategies. They are shifting from solely providing training compute power to higher-margin, more stable AI inference services, and building differentiated advantages by offering "data sovereignty" and controlled network connectivity services.