45k user reviews & 1 million votes, over 29k web hosting brands & 85k plans since 2004!
Menu
WHTop → 📋 News → AMD Helion Adds CPU Capacity for AI Agent Execution
Oct, 2026 : AMD Helion Adds CPU Capacity for AI Agent Execution
📅 - AMD has introduced the AMD Helion agentic AI rack, a configurable rack design for running AI agents that includes a hyperscale option with up to 32,768 processor cores. The blueprint separates agent execution from model inference in some deployments, giving cloud operators another infrastructure category to evaluate while leaving pricing, operating costs and production performance figures largely undisclosed.
For infrastructure buyers, the interesting question is how much ordinary server capacity sits around an AI model. Agents execute code, interact with applications and coordinate tasks between calls to inference services. Those activities create their own demands on processors and [...][... Check source for end of article ...]
VILNIUS, Lithuania, September 15, 2026. Melbicom, an EU-headquartered infrastructure operator, has opened its Marseille location for orders, with dedicated servers, Cloud VPS and custom hardware, alongside a new point of presence (PoP) for Premium and Volume CDN. The launch expands local infrastructure choice for customers serving Southern Europe, North Africa and the Middle East, with dedicated server ports up to 100 Gbps.
A Mediterranean Gateway to Three Continents
Marseille is a major interconnection hub for EMEA (Europ
📅 - Third Yandex Data Center Drone Attack in Russia Raises Cloud Recovery Questions - Yandex shut down its data center in Russia's Vladimir region on Sunday after a drone attack damaged infrastructure, the third strike affecting its facilities within days. The company reported no injuries and urged customers to activate disaster recovery plans as cloud operations entered emergency mode, leaving businesses to assess whether their remaining systems could support essential workloads.
For infrastructure customers, the immediate problem is execution. A recovery plan needs accessible backups, working credentials, available compute and a destination capable of running the application. If any of those depend on services affected by the same disruption, moving the workload becomes [...]
📅 - DE-CIX and TOWARDEX Bring New York Exchange Access to Boston - DE-CIX and TOWARDEX have connected Boston's Hub Express System to DE-CIX New York, offering regional networks access to peering and cloud services without renting space in traditional carrier hotels. The arrangement removes cross connect fees at both ends, according to the partners, changing part of the cost calculation for operators seeking connectivity beyond Boston's local network ecosystem.
For a regional internet provider or hosting operator, reaching an exchange can require more than purchasing a port. Equipment placement, transport, physical connections and maintenance arrangements all enter the bill. Removing the need for a separate colocation presence could make participation [...]
📅 - NVIDIA Opens Rubin Rack Designs to OCP as Power Demands Rise - NVIDIA contributed Vera Rubin rack and compute tray designs to the Open Compute Project on October 12, extending its MGX architecture with updated power, cooling and equipment interfaces. The changes give infrastructure suppliers a shared design foundation as large cloud providers and data center operators confront the electrical and mechanical demands of deploying increasingly dense AI systems.
For operators, the purchasing decision reaches well beyond the servers. A facility needs to deliver electricity, remove heat and let technicians service equipment without turning routine maintenance into a prolonged outage. Shared interfaces could reduce the amount of custom engineering between [...]
📅 - Oxide Raises $445M for Enterprise Owned Cloud Infrastructure - Oxide Computer Company has raised $445 million to expand production of cloud infrastructure that enterprises can own, as demand outstrips its manufacturing capacity. The Series D, led by Eclipse with participation from AMD Ventures, finances a hardware business confronting a question: how much cloud convenience can companies bring into their own data centers without importing new dependencies?
Buying infrastructure is straightforward enough. Making it behave like a cloud is the expensive part. Enterprises need provisioning, resource allocation, upgrades, fault handling and an operating model that developers will actually use. Otherwise, ownership delivers another equipment inventory and [...]
📅 - Lumen Technologies Partners With Eminence Grey on Private AI Deployments - Eminence Grey and Lumen Technologies have announced a partnership to combine dedicated artificial intelligence infrastructure with private connectivity across the United States, targeting government and enterprise workloads involving sensitive data. The companies are exploring infrastructure reservation options, leaving customers to establish which configurations are available, when they can be deployed and what operational controls the arrangement delivers.
For infrastructure buyers, the proposal brings compute and networking into the same purchasing discussion. An organization running sensitive AI workloads needs somewhere to process information, but also a controlled route between [...]
📅 - 
📅 - 
📅 - 
📅 -