SemiVoice

4 months ago

AMD's gaming revenue for PlayStation, Xbox, and Radeon fell 58% in 2024

➀ AMD achieved record revenue of $7.7 billion in Q4 2024 and $25.8 billion for the year, driven by data center products and Ryzen CPUs.

➁ AMD's Gaming revenue fell 59% year-over-year in Q4, with the segment becoming the smallest in AMD's portfolio.

➂ Despite the decline, AMD's Gaming segment remains profitable, but the focus is on Data Center profits.

AI AMD Data center EPYC Ryzen gaming revenue

4 months ago

ASUS AI POD with 36 NVIDIA Grace CPUs and 72 Blackwell GPUs shipping in March

➀ The ASUS AI POD, featuring the NVIDIA GB200 NVL72 platform, is set for March 2025 shipping.

➁ It boasts 26 NVIDIA GRACE CPUs, 72 Blackwell GPUs, and advanced cooling, offering significant AI performance enhancements.

➂ Designed for data centers, it supports large-scale AI training and inference, with ASUS providing full-service solutions.

AI ASUS CPU Data center GPU HPC NVIDIA cooling

AI Chip Industry Data center NVIDIA Supply chain

5 months ago

Phison's new Pascari D205V Gen5 data center SSD: up to 122.88TB, up to 14.6GB/sec reads

➀ Phison introduces the Pascari D205V SSD with a massive 122.88TB capacity and read speeds up to 14.6GB/sec; ➁ The SSD leverages Phison's X2 controller and 2Tb 3D QLC technology for superior performance; ➂ The SSD is designed to meet AI data needs and enhance scalability and power efficiency.

3D QLC Data center Performance Phison SSD Storage

trendforce

5 months ago

[News] NVIDIA Supplier Advantest: AI Phones May Sustain Chip Industry amid Declining Data Center Investment

➀ Major tech companies have significantly increased capital expenditures in recent years to acquire AI chips and build data centers. ➁ Global investment in data centers has recently slowed, raising concerns about the sustainability of the AI boom. ➂ AI phones may sustain the chip industry amid the decline in data center investment.

AI Data center Ethernet Interconnect Synopsys UALink Ultra Ethernet

5 months ago

Ultra Ethernet and UALink IP solutions scale AI clusters

➀ The growing demand for AI infrastructure requires scalable interconnects; ➁ Ultra Ethernet and UALink address the challenges of scaling AI acceleration clusters; ➂ Synopsys' IP solutions offer efficient and power-efficient interconnects for data centers.

6 months ago

Fujitsu's new 144-core Monaka chip: Arm, 2nm + 5nm chiplets, 3D-stacked CPU cores over memory

➀ Fujitsu unveils its Monaka processor, a 144-core Armv9-based chip designed for future data centers; ➁ The chip is built on TSMC's N2 process and features a CoWoS system-in-package with SRAM tiles and hybrid copper bonding; ➂ Fujitsu aims for superior energy efficiency by 2026-2027, using air cooling.

2nm 3D-stacking 5nm AMD EPYC Arm CPU Chiplets Data center Energy efficiency Fujitsu Intel Xeon Monaka

AI Bees Data center Energy Solutions Environmental Issues Meta Nuclear Power Regulatory Challenges

Meta's plan for a new nuclear-powered AI datacenter was foiled by bees

➀ Meta's plan to build a nuclear-powered data center was canceled due to the discovery of a rare bee species; ➁ The project faced environmental and regulatory issues; ➂ Despite the setback, Meta is looking for a new location and clean energy sources.

Analog Bits Builds a Road to the Future at TSMC OIP

➀ Analog Bits showcased its on-die sensing IP and power management solutions at TSMC OIP; ➁ The company highlighted its progress in 3nm and 2nm technology nodes; ➂ Collaborations with Arm were discussed, focusing on power management and clocking IPs.

2nm 3nm Analog Bits Arm Chiplet Data center EUV IP Development LDO Regulator Neoverse Power Management Sensing TSMC

AI servers Data center Exascale Compute Liquid Cooling NVIDIA SuperMicro

SuperMicro unveils NVIDIA GB200 NVL72 SuperCluster: liquid-cooled AI servers

➀ SuperMicro unveils its new NVIDIA GB200 NVL72 SuperCluster, a liquid-cooled Exascale Compute system in a single rack; ➁ The system includes 72 NVIDIA Blackwell B200 GPUs and offers high-speed interconnects; ➂ Supermicro provides end-to-end liquid-cooling solutions and comprehensive deployment services.

Dell PowerEdge XE9712: NVIDIA GB200 NVL72-based AI GPU cluster for LLM training, inference

➀ Dell has launched the new PowerEdge XE9712 with NVIDIA GB200 NVL72 AI servers, offering 30x faster real-time LLM performance over the H100 AI GPU; ➁ The system features 72 x B200 AI GPUs connected with NVLink technology, providing lightning-fast connectivity; ➂ Dell highlights the liquid-cooled system for maximizing datacenter power utilization and rapid deployment of AI clusters.

AI Data center Dell GPU LLM NVIDIA Performance Training inference

AI AI Infrastructure Chiplets Data center Optical Connectivity

8 months ago

How AI is Redefining Data Center Infrastructure: Key Innovations for the Future

➀ AI is transforming data center infrastructure with innovations like advanced connectivity and chiplets; ➁ The shift to optical connectivity is enhancing performance and efficiency; ➂ Chiplet ecosystems are enabling scalable and cost-effective solutions for AI workloads.

AI GPUs Data center HBM Rambus memory controller

8 months ago

Rambus details HBM4 memory controller: up to 10Gb/s, 2.56TB/sec bandwidth, 64GB per stack

➀ Rambus details its new HBM4 memory controller with speeds up to 10Gb/s and bandwidth of 2.56TB/sec; ➁ HBM4 offers 33% faster bandwidth than HBM3E and twice the speed of HBM3; ➂ SK hynix and Samsung are expected to mass produce HBM4 memory by the end of 2025.

AI server Data center GB200 NVL72 NVIDIA power consumption

8 months ago

NVIDIA's new GB200 NVL72 AI server: 'highest-power-consuming server in HISTORY' with 132kW TDP

➀ NVIDIA's new GB200 NVL72 AI server faces significant challenges due to its 132kW TDP; ➁ This makes it the highest-power-consuming server in history; ➂ The development of the server is experiencing delays due to technical issues.

AI chip Data center NVIDIA

9 months ago

This data center AI chip roadmap shows NVIDIA will dominate far into 2027 and beyond

➀ NVIDIA's dominance in AI GPU market extends to 2027 with Rubin Ultra AI GPU featuring 576GB of HBM4 memory; ➁ Competitors include AMD, Intel, Google, and others with their respective AI chip developments; ➂ TSMC's advanced CoWoS-L packaging technology and N3 process node support NVIDIA's next-gen AI GPUs.

anandtech

Kioxia Demonstrates Optical Interface SSDs for Data Centers

➀ Kioxia demonstrated an optical interface SSD at FMS, developed with funding from NEDO to reduce data center power consumption by up to 40%. ➁ The optical interface allows for data transfer over long distances, enabling storage to be kept in separate rooms with minimal cooling requirements. ➂ The demonstration showed a slight loss in IOPS performance but significant latency advantages over traditional copper network links.

Data center KIOXIA SSD

Data center NVIDIA SEMICONDUCTOR

A Post-AI-ROI-Panic Overview of the Data Center Processing Market

➀ The data center processing market shows robust growth, especially in processing, with a 20.3% increase. ➁ Nvidia dominates the market in terms of both revenue and profits, with a significant lead over competitors. ➂ The supply chain for AI GPUs has evolved, with memory supply becoming a critical component.

Mitigating AI Data Bottlenecks with PCIe 7.0

➀ AI data bottlenecks are addressed with PCIe 7.0, offering higher bandwidth and lower latency. ➁ PCIe 7.0 supports system-level optimization and reduces data center power consumption. ➂ The integration of photonics into PCIe ensures its relevance and benefits for future AI advancements.

AI Data center PCIe

pcwatch

Micron Announces World's Fastest PCIe 5.0 Data Center SSDs, Ideal for AI

1. Micron Technology introduces the Micron 9550 NVMe SSD series, boasting the world's fastest PCIe 5.0 data center SSDs; 2. These SSDs offer maximum sequential read speeds of 14GB/s and write speeds of 10GB/s, with up to 67% higher performance than competitors; 3. They are designed for AI workloads, with random read speeds of up to 3.3 million IOPS and write speeds of 400,000 IOPS, making them suitable for large language models and graph neural network training.

AI Data center SSD