Category
AI data & storage
-
The KV Cache is Becoming a Data Centre Asset, Not a Temporary Buffer
As AI inference moves towards longer conversations, larger context windows and more complex workloads, the information retained between interactions is creating a new infrastructure requirement. Data centre storage was largely discussed in familiar terms: capacity, performance, latency, resilience and cost per terabyte – AI inference is complicating that picture. Large language models generate and reuse…
-
Designing the AI Factory From the Data Out
Stacking more GPUs won’t solve the infrastructure crisis if your data-movement architecture can’t keep pace. In this deep dive, we explore why modern operators must stop designing data centres around physical rack space and start building them from the data out. Discover how architectural shifts from pioneers like Everpure are helping hyperscalers reclaim stranded power,…
-
AI Agents Are Changing the Data Centre Equation – And More GPUs Won’t Be Enough
As NVIDIA and Meta expand their infrastructure pact to power autonomous AI agents, the data centre industry faces a critical transition. The next competitive bottleneck is moving away from raw GPU volume and shifting toward system-wide utilization, sub-millimetre optical density, and complex thermal mechanics. Read why the rise of agentic inference means the individual chip…
-
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.
-
Everpure Unveils Bold Hyperscaler-Driven Growth Outlook
Everpure, the data storage company formerly known as Pure Storage, sent its stock soaring after unveiling a fiscal 2028 outlook that projects revenue growth of up to 45%, driven by new bets on AI workloads, hyperscaler partnerships and enterprise data software a sign that storage infrastructure is becoming one of the AI data center buildout’s…
-
VAST Data and CrowdStrike Alliance Redefines Data Center Security
As enterprise artificial intelligence workloads mature, protecting the model is no longer enough. The landmark integration between VAST Data and CrowdStrike shifts the defensive posture from physical and endpoint perimeters straight into the data repositories feeding high-density training clusters
-
Alibaba’s New Chip Lands Days Before Trump and Xi Meet — is that a Coincidence?
Alibaba unveiled the Zhenwu M890, its most powerful in-house AI chip yet, alongside a roadmap to match Nvidia’s annual upgrade cadence timed to land just ahead of a Trump-Xi summit.
-
Micron launches new Memory Module
Micron demonstrated a 512GB DDR5 RDIMM — enabling up to 12TB of memory in a single dual-socket server — with AMD and Intel both actively validating it for next-generation AI infrastructure
-
Frontier AI Model Learned to Hack Like an Expert – alone.
Frontier AI models are becoming capable of sophisticated autonomous cyberattacks, raising a new operational risk for data centres using agentic AI to manage power, security and physical infrastructure.
-
DRAM Buffer Vanishes as HBM4 Eats Capacity
HBM4 is consuming more wafer capacity as AI demand accelerates, pushing DRAM inventories to unusually low levels. For data-centre operators, the squeeze could extend beyond GPUs to server memory and enterprise storage.