Gigawatt campuses, power grids, cooling and the trillions being poured into AI compute buildouts.
516 articles
NVIDIA has introduced the Spectrum-6 networking platform, specifically designed to handle the massive data requirements of gigascale AI factories. This hardware aims to serve as a critical multiplier for computing power in training frontier models and powering agentic AI.
NVIDIA Blog · 60d ago
Meta is reportedly in talks to partner with Anthropic for a massive data center infrastructure deal. This move would mark a significant expansion of Meta's business model into providing compute resources for other AI labs.
Engadget · 64d ago
Former AWS senior executive Dave Brown is joining Meta to lead data center infrastructure efforts. This move fuels speculation that Meta is building its own cloud platform to compete with AWS.
GeekWire · 64d ago
Meta is reportedly in negotiations to lease its massive computing infrastructure to Anthropic. This potential $10 billion deal highlights the extreme scarcity of GPU resources required for large-scale AI model training.
New York Times Technology · 64d ago
Goldman Sachs reported record investment banking fees, driven by massive financing demands for AI infrastructure. Executives describe the current trend as an 'AI CapEx super cycle'.
The Next Web (TNW) · 65d ago
Nvidia is evolving from a GPU hardware provider to the chief architect of AI factories. This 'five-layer cake' strategy represents a full-stack integration of their computing ecosystem.
钛媒体 · 65d ago
Apple is reportedly looking to acquire AI chip companies to bolster its server infrastructure. The current M2 Ultra-powered servers are allegedly insufficient for the company's growing AI demands.
Engadget · 66d ago
Fujitsu has announced plans to manufacture high-end AI servers domestically in Japan, specifically designed to support NVIDIA's upcoming 'Rubin' GPU architecture. This initiative aims to address the growing demand for sovereign AI infrastructure and on-premises generative AI solutions.
ITmedia AI+ (日本) · 67d ago
Intel is investing 5 billion euros to expand its Leixlip manufacturing facility in Ireland. The investment focuses on upgrading production lines for Intel 3 process technology to meet the surging demand for Xeon 6 and next-gen AI server processors.
cnBeta (Full RSS) · 67d ago
Lazard's global head of power and infrastructure warns that energy demand for AI data centers remains a significant, unsolved challenge. The surge in demand is expected to exert upward pressure on natural gas prices.
Bloomberg Technology · 68d ago
Sugon has launched the Dawn 8000, China's first fully domestic super-AI fusion cluster capable of supporting 100,000 cards. The system integrates Hygon processors and connects directly to the National Supercomputing Internet.
Pandaily · 68d ago
Innolight addressed market concerns regarding 1.6T demand, stating that 2027 demand remains strong. While some structural adjustments exist, new cloud and AI model customers have increased their requirements.
36氪 · 68d ago
Shenzhen-based Cloud Leopard Intelligent has filed for a ChiNext IPO, aiming to become the first domestic DPU stock. The company is backed by Tencent as its largest shareholder and focuses on high-performance DPU SoC chips for AI computing.
36氪 · 69d ago
China has successfully launched its first 100,000-card computing cluster, marking a significant milestone in domestic AI infrastructure. The cluster has already successfully executed over 300 different AI applications.
量子位 · 70d ago
China has officially launched its first computing cluster featuring 100,000 cards, marking the arrival of the '100,000-card era'. The infrastructure is built entirely on domestic hardware, supporting full-precision computing from FP64 to INT8.
量子位 · 70d ago
Meta Platforms Inc. is investing $10 billion to build its first data center in Canada.
Bloomberg Technology · 73d ago
Investment bank Piper Sandler projects that Oracle Cloud Infrastructure (OCI) revenue will exceed market expectations by fiscal year 2027. The growth is driven by the efficient conversion of large-scale capital expenditures into actual production capacity and revenue.
36氪 · 73d ago
IBM has announced the upcoming release of its new z17 and LinuxONE 5 systems. These systems are designed to address critical data center constraints regarding space and operational costs.
36氪 · 74d ago
Nscale has closed a $900 million revolving credit facility to accelerate the construction of data centers across the US, Europe, and Asia-Pacific. This funding follows a significant Series C round earlier this year.
The Next Web (TNW) · 74d ago
Reports suggested that Nvidia's next-gen Kyber NVL144 architecture faced manufacturing setbacks with PCBs, potentially delaying the launch. Nvidia has officially responded, stating that their product roadmap remains unchanged.
cnBeta (Full RSS) · 74d ago
Nvidia's core energy partners, including Delta and ABB, report that the 800V high-voltage DC power solution is progressing as planned. Despite rumors of delays, the industry emphasizes that the pace is dictated by the maturation of the entire energy infrastructure supply chain.
36氪 · 75d ago
Nvidia's flagship Kyber AI server rack, designed for Rubin Ultra chips, has been delayed from 2027 to 2028. The delay is attributed to technical issues with a specific circuit board.
The Next Web (TNW) · 75d ago
NVIDIA's next-gen Kyber NVL144 rack architecture for Rubin Ultra chips is delayed from 2027 to 2028. The delay is attributed to technical manufacturing challenges with critical circuit boards.
cnBeta (Full RSS) · 75d ago
Nvidia's next-generation AI server rack systems are reportedly delayed by over a year due to manufacturing challenges. This setback has triggered a decline in Asian technology stocks linked to the AI supply chain.
Bloomberg Technology · 75d ago
Seagate executive Jason Feist discusses how AI workloads are redefining data center infrastructure, moving from a compute-first to a storage-centric architecture. The surge in AI-generated data is making high-capacity HDD and object storage essential for long-term data retention.
虎嗅 · 75d ago
Montage Technology's second-generation MRDIMM memory modules have reached 12800MT/s, doubling the performance of current mainstream RDIMMs. This technology is designed to address memory bandwidth bottlenecks in AI and high-performance computing applications.
IT之家 · 77d ago
Yunbao Smart, a DPU (Data Processing Unit) startup founded by a 24-year-old Stanford talent, is preparing for an IPO on the ChiNext board. Tencent holds a significant stake, highlighting the strategic importance of DPU infrastructure in the AI era.
钛媒体 · 77d ago
Extreme heat waves and the explosive growth of AI data centers are pushing US power grids to their limits. Major grid operators like PJM are facing record-breaking demand, leading to emergency measures and warnings of potential future power shortages.
虎嗅 · 78d ago
Meta developed a custom CXL chip called Vistara to integrate decommissioned RAM into new server architectures. This approach mitigates memory shortages and rising hardware costs without compromising performance.
Computerworld · 78d ago
Meta is shifting its strategic focus from model-level development to infrastructure. This move reflects the broader industry trend where AI value is migrating toward the underlying compute and data center stack.
钛媒体 · 78d ago