From the archive
AI hardware provides the computing power, memory and networks used to train and run artificial intelligence models. Follow news on GPUs, AI accelerators, data centres, chip design and energy demand. Read how changes in infrastructure affect performance, availability and cost, and distinguish measured results from vendor claims about what the next generation of computing will deliver.
Lead story
AWS’s JavaScript SDK v3.1143.0 adds support for Amazon Bedrock Agentic Retrieve to use the Bedrock Mantle (OpenAI Responses) endpoint, alongside changes to OpenSearch, security tooling and other AWS services. For developers, the release is a useful bundle of new integration options, not just another version number looking for…
Read the report & discussion ↗Watch Desk
Nvidia and AMD are reportedly pressing the Trump administration to ask Congress to change China-related chip export restrictions in the annual defence policy bill. The dispute concerns who sets the limits on advanced AI-chip sales, and how far they should reach.
Watch Desk
Investors are financing power plants, fuel cells and other on-site systems to get electricity to AI data centres faster than the grid can deliver it. The deals bring new money to a key AI bottleneck, but can tie data-centre operators to long contracts even as computing needs change.
Watch Desk
A Hugging Face community article argues that powerful AI agents could make compute costs balloon, even as model access looks remarkably cheap to users. Its worked example puts a month of rented GPUs for a large open-weight coding model at about $5,172 to $6,034, against the author’s roughly…
AMD Watch
AMD and Schneider Electric have produced a reference design for data centres built around AMD’s Helios AI racks, with configurations reaching 246kW per rack. The plans give operators a detailed starting point for the power and liquid-cooling demands of dense AI clusters, rather than leaving every project to…
Microsoft AI Watch
Microsoft has published a draft code of conduct for some of its AI products, with a final version planned for 2027. The principles include keeping AI under human control, preventing it from hiding actions and ensuring people can shut it down. The harder question is how those principles…
AWS AI Watch
Amazon and Qualcomm are working on custom AI inference chips and optical links for AWS data centres, joining compute and networking in one multi-generation collaboration. A Qualcomm filing also describes an Amazon-linked warrant, but its $60bn threshold is not a promise to spend $60bn.
NVIDIA Watch
Greenpeace says NVIDIA’s disclosed supply-chain emissions have risen 725% since 2020, and argues the chipmaker’s climate messaging gives too little attention to the footprint of making and using its products. The claims sharpen a practical question for the AI build-out: are efficiency promises keeping pace with its growing…
Watch Desk
AI data-centre expansion is becoming a campaign issue in the US, with voters weighing its promised investment against pressure on power, water and local communities. A September poll cited by the National Catholic Reporter found 58% of voters expected new data centres to have a mostly negative effect…
TSMC Watch
TSMC executives described a semiconductor industry growing faster than earlier forecasts, with AI demand driving a striking share of the expansion. The figures also point to pressure on chip design, memory and power efficiency, not just a race to build more processors.
NVIDIA Watch
NVIDIA is discussing insurance arrangements that could help lenders manage the risk of financing smaller cloud operators buying its chips, the Financial Times reports. The talks are at an early stage and may not produce a deal, but they point to a new question in the AI boom:…
ASML Watch
ASML plans to raise production capacity for its current-generation EUV systems by about 30% in 2027, with another increase of roughly 30% under consideration for 2028. That matters because making more advanced AI chips depends not just on chipmakers, but on the specialised machines that print their circuits.
AWS AI Watch
AWS has announced a set of updates spanning AI-agent monitoring, model inference and agent development. The most practical change for teams running language models is a SageMaker routing layer that AWS says can cut first-token latency by up to 82% in some scenarios.
ASML Watch
ASML makes the advanced lithography machines needed to manufacture many of the chips powering AI systems. That puts the Dutch company further upstream than the better-known chip designers and foundries, with a supply chain that is anything but easy to copy.
Hugging Face Watch
Before Nvidia agreed to buy Hugging Face, OpenAI offered to invest about $100 million and AMD and Salesforce also explored deals, CNBC reports. The detail shows how fiercely sought-after the open-source AI platform had become, and why control of its developer ecosystem matters.
Watch Desk
AI inference provider Modal Labs is nearing a $750 million funding round that would value it at $15.75 billion, according to TechCrunch. The reported terms show investors piling into the computing infrastructure needed to run AI, while leaving the round’s final terms unconfirmed.
Watch Desk
A Singapore startup says its ScienceGuru system led two public tests of AI-assisted research and posted the fastest reported times on two others in September. The results suggest that automated AI research is becoming measurable in public benchmarks, although two of the four headline results remain self-reported and…
Cerebras Watch
Cerebras and Gimlet Labs are combining wafer-scale processors with GPUs to build an AI inference cloud, with speeds of up to 3,000 tokens per second as their target. If delivered, it could make voice, video and agent-based AI feel markedly more responsive, rather than merely faster on a…
