GPU– tag –
-
TSMC’s CoWoS Capacity Dramatically Expands, Eliminating Supply Constraint for NVIDIA B300 Servers and Cutting Lead Times to 8-16 Weeks
Contrary Research Taiwan Overview TSMC's CoWoS (Chip-on-Wafer-on-Substrate) packaging capacity has significantly expanded, effectively resolving a major GPU procurement constraint that had persisted for over six months in 2025. As a resu... -
SK hynix Ships 12-Layer HBM4E Samples with Advanced MR-MUF Packaging, Achieving Over 20% Power Efficiency and 17% Thermal Resistance Improvement Over HBM4
Advanced Packaging News South Korea Overview SK hynix announced the sampling of its 12-layer HBM4E memory for next-generation AI systems, leveraging proprietary Advanced MR-MUF packaging. This innovation delivers 48GB capacity per stack ... -
Morgan Stanley Raises CoWoS Demand Forecast to 2.69 Million Wafers by 2027, Citing CPUs and ASICs as New Drivers
Industry Analysis / Moomoo Unknown Overview Morgan Stanley has significantly raised its demand forecast for TSMC's CoWoS packaging, projecting it to reach 2.69 million wafers annually by 2027. CPUs and ASICs are emerging as new key drive... -
Lumentum Secures Five-Year Order Backlog Amid Surging AI Optical Demand; NVIDIA Strategic Investment Strengthens Supply
HyperAI USA Overview Lumentum Holdings is experiencing a significant increase in demand for AI data center infrastructure, securing a remarkable five-year order backlog. This is driven by the industry's shift from copper to optical solut... -
Ayar Labs Joins NVIDIA NVLink Fusion Ecosystem to Deploy Co-Packaged Optics in Rack-Scale AI Infrastructure
Forge Global USA Overview Ayar Labs, an optical I/O data platform, announced its participation in NVIDIA's NVLink Fusion ecosystem in June 2026 to bring Co-Packaged Optics (CPO) technology to rack-scale AI infrastructure. The company aim... -
Quandela Validates Ultra-Low Latency Photonic QPU Integration with NVIDIA GPU Infrastructure via NVQLink
Quantum Computing Report France Overview Quandela has validated the ultra-low latency integration of its photonic Quantum Processing Unit (QPU) with NVIDIA's NVQLink infrastructure, establishing a real-time hybrid computing pipeline wher... -
Coherent and NVIDIA Scale Data Center Connectivity: Texas InP Fab Quadruples Wafer Production with $2B NVIDIA Investment
[News/Blog] USA Overview Coherent is expanding its indium phosphide (InP) wafer fab in Sherman, Texas, with a major $2 billion investment from NVIDIA to quadruple wafer production capacity and secure supply for AI infrastructure. This ex... -
SuperX to Unveil 1.6T Optical Modules for AI Data Centers at Interop Tokyo 2026
The Data Center Engineer Japan Overview SuperX is set to debut its 1.6T optical module solution at Interop Tokyo 2026, designed to address the demanding bandwidth requirements of large-scale AI training and inference networks. The soluti... -
TSMC Accelerates Advanced Packaging Strategy: Addressing CoWoS Capacity Shortfall, Driving AI Accelerator Demand
Umbrex Taiwan Overview TSMC's strategy is founded on advanced process technology, manufacturing yield, reliable execution, and a broad design ecosystem. As chip architectures shift towards chiplets and 3D integration, advanced packaging ... -
AI Computing Drives 1.6T Optical Modules, CPO Slashes Power Consumption by Up to 68%
EqualOcean China Overview With Moore's Law slowing, horizontal scaling of AI clusters is unavoidable, pushing electrical interconnects to bandwidth, latency, and power limits. By 2026, the shift from copper to optics is a defining data c...
