45k user reviews & 1 million votes, over 29k web hosting brands & 85k plans since 2004!
Menu
WHTop → 📋 News → Huawei: Agentic AI May Drive 100-Fold Compute Cluster Growth
Huawei: Agentic AI May Drive 100-Fold Compute Cluster Growth
📅 - Huawei expects AI agents to drive a 100,000-fold increase in global token consumption by 2035, a projection that would require compute clusters to expand dramatically while data center operators wrestle with power, cooling and network limits. The company estimates agents could generate more than 90 percent of traffic, making machine-to-machine activity a dominant infrastructure workload within a decade.
That is a very large number attached to a very distant forecast. It also lands at a moment when cloud providers, chipmakers and data center developers are already spending heavily on infrastructure whose utilization economics remain difficult to predict.
Huawei's central assumption is [...][... Check source for end of article ...]
📅 - Gartner: Power Constraints Draw Enterprises Into AI Power Supply - Enterprise energy assets are moving closer to the center of AI infrastructure economics as surging data center electricity demand forces companies to rethink who owns, controls, and sells power. Gartner expects enterprise-owned energy assets to reach $10 trillion by 2030, creating a market where large companies could supply electricity directly to grids and AI data centers globally.
The forecast points toward an awkward convergence for infrastructure operators. Data center capacity can be designed, financed and filled with processors faster than adequate electrical capacity can necessarily be connected. That has already changed site selection in several major markets, where megawatts [...]
📅 - GTZHost Publishes Technical Guide on Resolving GPU VRAM Memory Leaks and OOM Errors Without System Reboots - As enterprises continue to scale their Artificial Intelligence and Large Language Model (LLM) inference pipelines, maximizing hardware uptime has become a critical operational priority. One of the most frequent issues encountered by data scientists and DevOps engineers is the persistent "Out of Memory" (OOM) error, where GPU Video RAM (VRAM) remains fully allocated even after a computational script has crashed or terminated. To address this widespread issue, GTZHost has published a comprehensive technical guide detailing how to clear stuck VRAM without requiring a highly disruptive system reboot.The Architecture of Stuck VRAM The tutorial begins by diagnosing
📅 - GTZHost Releases Comprehensive Guide on GPU Observability and Monitoring Stacks for Enterprise AI - As enterprise adoption of Artificial Intelligence and Large Language Models (LLMs) accelerates, the operational complexity of managing the underlying hardware has skyrocketed. To address the critical visibility gap in high-performance computing, GTZHost has published an in-depth industry guide detailing the modern monitoring stacks required for production-grade GPU environments.The Limitations of Traditional Command-Line Diagnostics The publication emphasizes that "GPU monitoring" is not a single tool, but rather a tiered software stack. Historically, system administrators relied heavily on basic command-line utilities like nvidia-smi. While usefu
📅 - New Jersey to Brean Cable Adds 24 Fiber Pairs Across Atlantic - EXA Infrastructure will build a subsea cable spanning 6,552 kilometers from New Jersey to Brean, United Kingdom, creating a new transatlantic route that will feed traffic onward to Slough and continental Europe. Due in late 2029, EXA Meridian is designed with 24 fiber pairs and more than 500 Tbps of capacity for hyperscalers, neoclouds and other large bandwidth buyers.
The route is the more interesting part of the project. Transatlantic capacity is plentiful in aggregate, but large buyers do not purchase networks by looking only at total terabits. They care about where cables land, which terrestrial paths carry traffic away from the beach, how much infrastructure overlaps with existing [...]
📅 - Aging Branch Networks Drive New Wave of SASE Spending - The SASE market reached $3.5 billion in the second quarter of 2026 as enterprise spending on SD-WAN accelerated alongside broader security services, extending annual growth above 20 percent for a fifth straight quarter, according to Dell'Oro Group. The shift points to a branch-network refresh cycle in which aging routers, cloud-delivered controls, and rising capacity requirements are colliding across large estates globally.
For infrastructure buyers, the more useful signal sits beneath the growth rate. Enterprises are spending again on branch networking, but the pattern is selective rather than synchronized. Equipment approaching support limits, sites carrying more traffic, and locations [...]
📅 - Nebius Raises GPU Cloud Rates Again as AI Demand Accelerates - Nebius will raise on demand AI cloud prices again from October 1, increasing rates for selected NVIDIA GPUs by 17% to 21% while also lifting some CPU and memory charges. The second increase in three months suggests scarce accelerator capacity is giving specialist cloud providers more pricing power, even as customers face higher infrastructure costs per hour.
The immediate problem for enterprise buyers is budgeting. NVIDIA H100 capacity will move from $3.85 to $4.50 per GPU hour, while H200 pricing rises from $4.50 to $5.40. NVIDIA B200 capacity moves from $7.15 to $8.50, and the newer B300 goes from $7.85 to $9.50. Those increases land on workloads whose economics are already difficult to [...]
📅 - 
📅 -