GPU– tag –
-
AI Chip Packaging Bottleneck Intensifies: NVIDIA Explores Intel EMIB Amid TSMC CoWoS Shortage, Amkor Commits $7B to Arizona
The Economy Japan Overview The burgeoning AI chip market faces a critical bottleneck in advanced packaging, particularly with TSMC's CoWoS technology, prompting NVIDIA to reportedly consider Intel's EMIB for future GPU production. This s... -
Nvidia Invests Billions in Silicon Photonics for Rubin & Blackwell GPUs, Aiming for Petabit-Scale Bandwidth with Minimal Heat
Medium USA Overview Silicon photonics is shifting its focus from long-haul links to short-reach inter-chip interconnects within data centers and AI accelerators, addressing the 'memory wall' problem. Nvidia is investing billions in photo... -
Lightmatter Reports Optical Interconnects Achieve 100x Data Transfer Speed, Up to 5x Power Efficiency as AI Industry Races to Scale Integration
Tom's Hardware USA Overview As copper wiring approaches its physical limits, photonic interconnects are rapidly emerging as an innovative solution for AI workload data centers. Lightmatter's CEO states that optical interconnects can boos... -
Photonics Achieves Bandwidth, Latency, and Power Efficiency Breakthroughs, Attracting Billions in New Investments for AI Infrastructure
International Banker International Overview Photonics is attracting billions of dollars in new investment due to its essential role in AI infrastructure, achieving groundbreaking breakthroughs in bandwidth, latency, and energy efficiency... -
TSMC Unveils Next-Gen ‘COUPE’ Silicon Photonics Platform for AI Optical Interconnect, Targeting 1.6 Tb/s Optical Engines by 2026
Tech Investments Taiwan Overview TSMC has announced 'COUPE,' a silicon photonics platform for co-packaged optics (CPO) in the AI era. COUPE innovatively stacks two dies—a photonic IC and an electronic IC—using SoIC bonding, placing the o... -
YieldWerx Analyzes 2026 Photonics & Optical I/O Trends: In-Package Optical I/O Becomes Essential for AI Workloads
yieldWerx USA Overview YieldWerx has released an analysis of key trends and challenges in photonics and optical I/O for 2026, highlighting that optical interconnects are undergoing a dramatic shift from system-to-system to in-package com... -
Lightmatter’s Photonic Processor Achieves ~10x AI Workload Speedup and 3.5-5x Power Efficiency Over GPUs
Daily Maverick USA Overview Lightmatter's photonic processor has demonstrated the ability to execute AI workloads approximately 10 times faster than traditional GPUs, while also improving power efficiency by 3.5 to 5 times. This achievem... -
STMicroelectronics Launches 300mm Wafer Production of PIC100 Silicon Photonics for AI Data Centers
EE Times International Overview STMicroelectronics has begun mass production of its PIC100 silicon photonics platform on 300mm wafers to meet the escalating demands of AI data centers. This platform integrates BiCMOS for laser drivers wi... -
AMD Unveils Instinct MI455X AI Accelerator with 432GB HBM4 Memory, Surpassing Nvidia’s Rubin GPU, to Challenge Dominance in Data Centers with CDNA 5 and Helios Rack-Scale Architecture
Tom's Hardware USA Overview AMD has announced the Instinct MI455X AI accelerator, integrating its CDNA 5 and Helios rack-scale architectures. This new product features a substantial 432GB of HBM4 memory, significantly exceeding the 288GB... -
Nvidia Details Rubin Architecture Optimizations for AI Inference: Tackling Memory Bottlenecks and HBM Shortages Amid Intensifying AI Chip Race
Tom's Hardware USA Overview Nvidia has unveiled comprehensive inference optimization strategies for its next-generation Rubin AI accelerator architecture, focusing on efficient management of Mixture-of-Experts (MoE) models and improved d...
