AI Infrastructure & LLMs
Large language models, AI assistants, data centres, AI chips, cooling, optical networks and the power systems behind modern artificial intelligence.
-
AI’s Next Bottleneck Is Hiding Between the Chips
The AI race needs more than faster processors. A new Marvell–GlobalFoundries agreement puts the connections between them in the spotlight.
-
Nvidia’s New 84GB Graphics Card Makes Memory the Main Event
Nvidia’s RTX PRO 5500 puts a striking amount of memory on one card. The interesting part is what that extra room could let AI teams do—and what it does not prove about speed.
-
Perplexity’s Windows AI Agent Puts Your RTX PC to Work—With a 24GB Catch
Perplexity Portable Computer now supports Windows RTX PCs. Local workflows are the attraction; the 24GB GPU-memory requirement is the catch.
-
Nvidia and Google Want Data Centres That Know When to Turn Down the Power
A new alliance wants AI facilities to ease pressure on the grid by changing when they consume electricity. The promise depends on commitments utilities can verify.
-
Paper2Agent Turns Research Papers Into AI Tools That Can Do the Work
A new Nature study moves beyond chatting with PDFs: Paper2Agent packages research code into tools that an AI assistant can actually use.
-
Huawei’s Million-Processor Plan Takes the AI Fight Beyond Nvidia’s Chips
Huawei’s Peerium architecture targets million-processor systems. The important contest is shifting from individual chips to the machines built around them.
-
Claude’s Next Home Could Be a Giant Data Centre in Regional Queensland
Anthropic’s first Australian data-centre agreement puts regional Queensland on the AI map. A signed lease, however, is only the beginning of a very physical project.
-
Nvidia Is Using AI to Untangle the Supply Chain Behind AI
A fast GPU cannot fix a missing component. Nvidia and Palantir are bringing AI to the complicated chain of decisions that turns parts into working AI systems.
-
Nvidia’s Next AI Advantage Is Getting More From Every Watt
Nvidia and Lambda report 24% more AI output within the same power budget. The result points to a different way to expand computing capacity: manage electricity more intelligently.
-
AI’s Next Building Boom Starts With the Power Supply
SB Energy’s latest fundraising plan highlights a less glamorous part of the AI race: building the electricity supply and data centres together. The money is important. Delivering working infrastructure is the harder test.
-
Nvidia’s Quantum Computing Push Is Really a Software Power Play
Nvidia is expanding CUDA-Q into the difficult business of making quantum computers reliable. The immediate advance is in the tools researchers use, with much bigger ambitions behind it.
-
Europe Is Waking Up to the Risk of Renting Its AI Future
Christine Lagarde’s latest warning pushes AI dependence into the economic debate. Europe’s challenge is to build real alternatives while retaining the benefits of working with the rest of the world.