Tag
Microsoft highlights how its datacenters in Quincy have driven economic growth, reducing poverty rates, creating jobs, and funding local infrastructure.
This article reports statistics on global datacenter power usage, highlighting that the US accounts for 43% of the world's total, with varying percentages of electricity consumption dedicated to datacenters across different countries.
Arm announces new C2-Ultra CPU and G2-Ultra NX GPU IP cores for mobile devices and CSS N4 for datacenters, with technical improvements and comparisons to previous architectures.
A new arXiv paper explores using reinforcement learning to reduce AI datacenter energy consumption by dynamically controlling power for LLM training workloads across scales from a single GPU to entire fleets.
Opinion piece arguing that local AI models will never win because they are weaker, more expensive, and less efficient than datacenter inference, due to batching and GPU advantages.
AMD's Q2 2026 earnings show data center revenue surging 107% YoY to $6.7 billion on AI demand, while gaming revenue fell 31% due to component shortages and price hikes. Overall revenue hit a record $11.5 billion, up 50% YoY.
OpenAI announces Project Camellia, a large-scale datacenter in Effingham County, Georgia, with commitments to not raise residential electricity rates, minimal water use, $80 million in community benefits, and independent audits for accountability.
NVIDIA announces Vera Rubin platform, delivering 10x more throughput per megawatt than Blackwell and claiming lowest token cost for AI factories through extreme co-design across seven chips and five rack trays.
The article deeply deconstructs how CXL technology breaks the AI memory wall, analyzes memory bottlenecks from training to inference stages, and the application prospects of CXL in data centers.
Microsoft's 2026 sustainability report reveals a 25% increase in carbon emissions, driven by datacenter expansion and AI demands, and notes sustainability solutions are not keeping pace.
Nvidia's stock has fallen 15% from its peak as GPU shortage eases, while memory demand surges, making companies like Micron the new AI bottleneck.
Miles Brundage observes that while the capability gap between open and closed AI models is months, the gap between models runnable on datacenter versus personal hardware is years, which matters for emergency and privacy use cases.
Benjamin De Kraker critiques AI companies pivoting to datacenter provision, arguing that if AI inference is unprofitable, the demand for datacenters to run those models is also questionable, using a train analogy.
Shawn Wang announces taking over a new media lab in San Francisco as the home for engineer-creatives, a third place for making, and a finishing school for technical storytellers; the space came with a surprise datacenter rack.
SemiAnalysis refutes the widely circulated claim that half of 2026 US datacenter capacity is delayed or canceled, attributing it to inaccurate models and showing their own data indicates only minor changes in forecasts.
Anthropic CEO Dario Amodei said in an interview that we are approaching the end of the exponential curve, internal models can already complete 100% of coding tasks, and predicts a 90% probability of a 'country of geniuses in a datacenter' within 10 years.
This article questions the sustainability of large-scale AI datacenters, discussing water and energy demands, and evaluating potential solutions like orbital datacenters and efficiency improvements.
Nvidia unveiled its photonics co-packaged optics switch with Lambda, aiming to reduce power consumption and failure points in large GPU clusters for AI workloads.
xAI is increasingly acting as a datacentre REIT by renting compute capacity to Anthropic and Google, potentially recouping billions in capex within 18 months. The deals highlight the ongoing AI compute crunch and may serve strategic purposes beyond pure commerce.
A detailed guide on installing SmartOS on a datacenter machine without physical access or remote KVM, using VirtualBox to create a template ZPOOL image.