Category
Hardware
-
Nvidia Reports on Tuesday – Operators Should Read It Differently From Investors
Nvidia reports on Tuesday, and for data centre operators the results are a delivery schedule more than a scorecard. This piece explains what the Vera Rubin ramp, supply constraints stretching to 2028 and rising memory costs mean for AI infrastructure planning, and the three things operators should listen for.
-
The AI Data Centre Is Becoming a System – and Arm Wants to Help Design It
Arm’s 3nm AGI CPU launch shifts the AI data centre narrative from isolated silicon to integrated rack-scale architecture optimised for agentic workloads.
-
The Next AI Bottleneck Isn’t Compute. It’s Memory
While the industry remains fixated on raw compute, a silent bottleneck is stalling AI infrastructure. From the emergence of massive KV-cache workloads to the limits of HBM, the real challenge of scaling AI has shifted from how fast a chip can think to how fast data can reach it.
-
The GPU Is No Longer the Data Centre
Silicon chauvinism has blinded the tech industry to a looming crisis. For years, we treated the GPU as the sole monarch of AI infrastructure. But as power grids buckle and fluid dynamics rewrite server design, the individual chip is no longer the story—the entire data centre is the computer.
-
nVent Wants to Own the AI Rack
AI is changing the data-centre rack. As power and heat rise, vendors across the sector are integrating power, cooling and rack infrastructure to support denser compute.
-
Intel Targets the AI Inference Bottleneck as Data Center Workloads Evolve
Intel is positioning CPUs, inference accelerators, memory and networking as parts of one AI infrastructure system as data-centre workloads shift towards production inference.
-
HPE Juniper Takes AI Networking to Gigawatt Scale
HPE’s expanded Oracle collaboration shows why networking is becoming critical AI infrastructure, with GPU utilisation, congestion, telemetry and resilience now closely linked.
-
Arm Has Quietly Become One of AI’s Most Important Data Center Companies
Training runs and inference queries get the headlines, but neither happens without an orchestration layer; the CPU work of scheduling, memory management, and increasingly, coordinating fleets of autonomous AI agents rather than single prompts.
-
AMD Joins the Trillion Dollar Club
Advanced Micro Devices shares surged more than 9% to a record $613.31, pushing the company’s market capitalization above $1 trillion for the first time
-
Redefining the Edge: Why Nokia’s Strategy Matters to Data Center Operators
While hyperscalers focus on gigawatt-scale facilities, Nokia’s recent partnerships with NVIDIA and Microsoft point to a more distributed future. By pushing AI-RAN workloads and automated data structures to the network edge, the boundary of where AI compute actually executes is permanently changing.