Chip Capacity Constraints Put Brakes on AI Spending Growth
Advanced packaging and memory supply, not demand, now set the ceiling for AI infrastructure expansion.
Compute & Hardware
Chips, datacenters, interconnect, and the power they draw.
AI Workloads Expose Limits of Platforms Built for Pre-AI Era
A sponsored analysis argues that current compute and storage platforms were designed for traditional workloads, and AI's demands are now revealing their architectural shortcomings.
AMD Positions for Agentic AI Surge
AMD's hardware roadmap is now aligned with the growth of agentic AI workloads, a bet analysts expect to pay off.
Why Amazon Could Spin Off AWS and Annapurna Labs
A new analysis argues Amazon might separate AWS and its chip unit to unlock value and sharpen focus, though the idea remains speculative.
Microsoft's AI Hype Outpaces Its Revenue Impact
A new analysis argues that while AI dominates Microsoft's public narrative, it has yet to become the company's main business driver.
AMD Lays Out Rack-Scale AI Roadmaps to Capture AI Money
AMD's rack-scale system roadmaps signal a deliberate move from component vendor to AI platform provider.
Nvidia Turns AI Agents Loose on Chip Engineering
Nvidia is applying AI agents to its own chip engineering work to speed up the process, according to The Next Platform.
DoE Kicks Off $5B Genesis Push for AI Computing
The Department of Energy has signaled the start of a major, AI-focused computing initiative with a $5 billion commitment.
China's LineShine: An All-CPU Exaflops-Class Supercomputer
A deep dive from The Next Platform examines how China's LineShine supercomputer reaches exaflops-class performance without GPU accelerators.
HPE Pivots to Agentic AI to Reclaim Datacenter Relevance
The Next Platform reports that HPE is betting on agentic AI workloads to drive demand for its datacenter hardware and software.
HPE’s GenAI Push Gains Traction With Enterprises, Sovereigns, Neoclouds
HPE’s first major GenAI wave shows its AI infrastructure is winning demand beyond the usual hyperscale crowd.
AI's Real Bottleneck Isn't Chips—It's Electricians
A shortage of skilled electricians, not just GPUs and RAM, is now the limiting factor for AI infrastructure buildouts.
AI-Native Software Development Reshapes Compute Demands
The Next Platform examines how building software in the AI-native era changes the underlying compute and hardware landscape.
Arista Modular Switches Target Network Scale and Scale-Out
Arista's new modular switches aim to scale across entire networks while also supporting scale-out architectures, according to The Next Platform.
NextSilicon's Maverick-2 Dataflow Engine Takes On CPUs and GPUs
The Next Platform reports on a new HPC chip that uses a dataflow architecture to compete with mainstream processors.
HPC Simulations Reveal How Dinosaurs Moved and What It Means for Robotics
High-performance computing is enabling researchers to model dinosaur locomotion in unprecedented detail, with implications for biomechanics and robotics.
Oracle First in Line for AMD MI450 'Altair' GPUs and Helios Racks
Oracle's early access to AMD's next-generation MI450 GPUs and Helios racks signals its bet on AMD's AI infrastructure roadmap.
Atomically Thin Material Shrinks Qubits Without Sacrificing Performance
A new approach using 2D hexagonal boron nitride targets the twin challenges of quantum computing: miniaturization and qubit quality.