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
IBM says it has tuned NVIDIA Nemotron models for IBM Bob, its self-hosted AI coding assistant, giving organisations a way to run agentic software development on infrastructure they control. The practical draw is local deployment for sensitive or air-gapped environments, alongside IBM’s claim of up to four times…
Read the report & discussion ↗Watch Desk
Reka AI has introduced Rho-1, a 19-billion-parameter model built to handle text, images, video and robot-control actions in one network. The notable idea is a shared model and context for tasks that are often split across specialised systems, including work involving robots.
NVIDIA Watch
AM Intelligence says it has placed firm orders for 20,000 more Nvidia Rubin GPUs, to be deployed in India and Malaysia from the second quarter of 2027. The expansion would bring its committed Rubin capacity to about 29,000 GPUs and nearly 100 megawatts, extending a major buildout of…
Watch Desk
A VentureBeat survey puts Google platforms first among organisations running AI in production: 48% of respondents named Gemini or Google Cloud as their primary platform, against 22% for OpenAI and 13% for Microsoft Azure. The results offer a useful snapshot of enterprise choices, though the primary-platform figures come…
Watch Desk
HIVE Digital is converting former Bitcoin-mining sites into Canadian GPU cloud infrastructure, with two reported contracts worth an estimated $570 million combined. The plan links repurposed power-hungry facilities to demand for domestic AI computing, including a Cohere deployment.
Watch Desk
The AI boom’s benefits may be reaching ordinary Americans through construction jobs, tax revenue and consumer use, not just through the technology’s effect on office work. Noah Smith makes that case in a new essay, while acknowledging that data centres bring costs as well as gains.
Watch Desk
Tencent is reportedly set to rent about 100,000 AI chips from Oracle for five years, in a deal estimated at $7 billion. If accurate, it would be a major example of Chinese firms obtaining overseas AI computing capacity while access to advanced chips remains a live policy issue.
Watch Desk
Hitachi and Germany’s Agile Robots are partnering to develop physical AI systems for complex factory tasks, combining Hitachi’s edge AI technology with Agile’s robotics platform. The plan is to commercialise the resulting systems globally through Hitachi’s HMAX Industry portfolio, putting AI closer to machines and production lines rather…
NVIDIA Watch
NVIDIA has selected engineering firm Jacobs to deploy a digital twin at one of its largest US AI research and development facilities. The three-year project aims to bring live facility data and simulation together, with Jacobs describing a path towards more autonomous operations as AI data centres grow…
AMD Watch
AMD’s deals to supply AI chips to OpenAI and Meta include warrants for up to 320 million shares, and the share-price threshold for the final batches has now been crossed. But the shares are not a cheque already written: the warrants depend on the customers buying the chips,…
Intel AI Watch
Intel Neural Compressor 3.10 adds new quantisation options for PyTorch and Keras/JAX, alongside support for more models and hardware. For teams trying to fit AI workloads to available accelerators, the release offers practical new knobs, not just a version number and a ceremonial cake.
Watch Desk
Scotland is weighing at least 23 proposed hyperscale data centres, including one planned site with capacity comparable to a large power station. The argument is not just about the AI they could run: it is about electricity, water, noise and what local communities get in return.
Watch Desk
Hollywood is turning real tech founders into film subjects as public confidence in the industry frays. The new wave includes films about Sam Altman, Mark Zuckerberg and Elon Musk, alongside questions about AI’s impact on creative work.
Watch Desk
People can dislike the AI industry’s relentless push and still find chatbots useful enough to keep coming back. A new MIT Technology Review essay argues that public scepticism may be aimed less at AI itself than at the companies’ insistence that it is inevitable.
vLLM Watch
vLLM 0.31.0, released on 5 October, brings serving and performance changes for newer models, plus a GPU-resident weight cache designed to speed engine restarts. For operators, the gains come with several compatibility changes that make this one worth testing before upgrading production.
Submer Watch
Submer Group has launched Corenix, a company focused on modular data-centre platforms for AI operators, hyperscalers and neoclouds. The notable offer is an integrated route from factory testing to on-site commissioning, with Corenix saying it will set out a programme and go-live date before a contract is signed.
ASML Watch
A TrendForce report says Chinese fabs bought about 90 ASML immersion DUV lithography systems in each of 2024 and 2025, spending more than $13bn on the machines. The reported stockpile matters because these tools can support advanced chip production, including chips used in AI systems, even as export…
Submer Watch
Submer and Saudi digital infrastructure company ACES have signed a memorandum of understanding to jointly pursue high-density, AI-ready data centres in Saudi Arabia. The proposed partnership brings facility engineering, cooling, IT integration and potential investment under one arrangement, rather than leaving customers to assemble the pieces themselves.
OpenRouter Watch
OpenRouter has introduced two beta tools that let AI models run commands in a provider-hosted sandbox, including models from different companies. The practical draw is that developers can add code execution without building and maintaining the sandbox themselves, while retaining a route across supported models.
Watch Desk
Toshiba plans to double its production capacity for hard disk drives used in AI data centres by FY2027, from its 2025 level. The company is also targeting a 30% share of storage capacity, up from about 10%, according to Nikkei Asia.
