Hardware
AMD Unveils Threadripper Halo Station for On-Device AI Workloads
A liquid-cooled workstation aims to bring supercomputer power to developers without relying on the cloud.
YESTERDAY, STILL WORTH KNOWING
Still
Crunchy.
Hardware makers push local AI power while OpenAI's new reasoning technique unsettles safety experts.
Local AI Powerhouses
AMD has introduced the Threadripper Halo Station, a liquid-cooled workstation designed to run massive AI models directly on a device. The system is built around a 96-core Threadripper PRO 9995WX processor and can hold up to four Instinct MI350P accelerators. It offers up to 576 GB of HBM3E memory, which allows developers to process trillion-parameter models locally without needing cloud infrastructure. This addresses the growing demand for on-device processing and privacy.
Nvidia is also accelerating local AI at IFA 2026. The company announced a partnership with Microsoft to make it easier to run AI agents on personal hardware. This collaboration aims to simplify the setup of digital assistants on your own desk. Nvidia also unveiled its new N1X System on Chip (SoC) for the RTX Spark platform. This chip represents Nvidia's first entry into the Arm-based Copilot+ PC market, offering a balance of performance and efficiency for AI-focused laptops.
Hardware and Ecosystems
Acer has unveiled the Swift Blade 14, a new sub-brand aimed at making thin-and-light Windows laptops feel less fragile. The device weighs just 799 grams, making it one of the lightest 14-inch Windows laptops on the market. It features a premium OLED display and a slim design, though its thermal performance remains to be seen. The laptop will launch across Europe, the Middle East and Africa in December, with pricing yet to be announced.
In a major move for the AI community, Nvidia has agreed to acquire Hugging Face for $12.93 billion. The deal aims to scale the platform's infrastructure and expand access to AI tools for developers. This acquisition highlights the increasing value of software ecosystems in the AI hardware market and the growing importance of the AI community in driving innovation.
AI Safety and Security
OpenAI has launched GPT-6 Astra, a model that the company claims is a generational leap in capability for cybersecurity and software engineering. The launch coincides with a new reasoning technique called 'recurrent depth' that allows the model to operate outside standard sequential thinking. This shift has unsettled AI safety experts who worry the new approach may complicate safety measures designed to keep models in check.
Cloudflare has teamed up with OpenAI to automate the discovery and remediation of software vulnerabilities. The new feature for its Managed Defense service uses OpenAI's Daybreak models to analyse production traffic and security signals. The system identifies threats, proposes code patches, and prepares edge mitigations. This tool aims to help security teams prioritise critical issues and patch them faster by automating routine vulnerability management.