NVIDIA · Original source
NVIDIA has outlined how semiconductor manufacturers are using accelerated computing, physics-based AI, AI agents and digital twins to speed up chipmaking work. Examples include COMSOL and Applied Materials simulations running up to 10x faster with NVIDIA cuDSS, Siemens integrating cuLitho for up to 17x acceleration, Lam Research predicting plasma behaviour in milliseconds with PhysicsNeMo, and Advantest cutting model-selection times from hours to minutes.
OpenAI · Original source
OpenAI has published a customer story on how Cisco pairs its network telemetry with OpenAI models to detect and respond to security threats. Cisco President and Chief Product Officer Jeetu Patel says Daybreak and GPT-6 Astra have helped Cisco find and patch vulnerabilities more efficiently. Through the Daybreak Defense Network, the companies are exploring faster exposure triage and control enforcement in Cisco Cloud Control.
Fireworks AI · Original source
Fireworks AI unveiled new products at Forge, its first annual conference, aimed at helping companies build specialised AI on open models. Launches include Fireworks Ultra-fast on NVIDIA Blackwell GPUs, delivering up to 1,500 tokens per second in public preview, a half-price Background Mode, Nexus Desktop with Pro plans from $8 a month, and Fireworks Sovereign, which brings its inference engine to customers' own infrastructure.
OpenAI · Original source
OpenAI has appointed Jill Hazelbaker as Chief Corporate Affairs Officer, leading the company's global communications and corporate positioning. She joins from Uber, where she was President and Chief Corporate Affairs Officer for a decade. She previously led communications and public policy at Snap, held senior roles at Google, and was National Communications Director for John McCain's 2008 presidential campaign.
OpenAI · Original source
OpenAI says its models are helping engineers design, simulate and build physical products across mechanical, electrical and system design. Two team members used GPT-6 Astra to design a Codex-connected test device in under a week, though manufacturing still needed several review rounds. Model usage across OpenAI's hardware teams has grown 23-fold since February. OpenAI is expanding work with Chip Ganassi Racing and announced a research collaboration with Mazda.
OpenAI · Original source
OpenAI has published a case study on Heimstaden, a European residential property operator using ChatGPT Enterprise, Codex and OpenAI APIs across its business. Heimstaden says AI has cut the cost of virtually staged listing images in the Netherlands by more than 99%, half its employees adopted ChatGPT within two to three months, and an analyst used Codex to assess climate risk across 20,112 property assets. It is also building AI-driven procurement and property management systems.
Google · Original source
Google.org is providing more than $9.5 million in funding to five organisations building AI tools and training with and for educators. The recipients are The City University of New York, Fab AI, the Krause Center for Innovation, Teaching Lab and 1EdTech Global. The projects cover teacher computing literacy, open models for lesson planning and early reading, responsible AI education in school districts, real-time support for teachers and trusted learning standards.
NVIDIA · Original source
NVIDIA has outlined how cities and partners are using its physical and agentic AI tools for public services, infrastructure planning and shared local AI infrastructure. Examples include resident-service agents from 1MillionBot in Spain, vision AI work by K2K in Wrocław, Poland and ITMAX in Malaysia, digital twins from Linker Vision and MVRDV in Taiwan, and a sovereign AI factory at Morocco's Tanger Med port. NVIDIA is presenting at Smart City Expo World Congress in Barcelona, 3–5 November.
SambaNova · Original source
SambaNova says Supermicro will integrate, deliver and support inference systems that pair NVIDIA GPUs with SambaNova SN50 RDU chips, using GPUs for the prompt-processing phase and SN50s for token generation. The systems will roll out in phases, starting with an initial batch of 10,000 SN50 chips, going first to NeoClouds and sovereign clouds and later to enterprise data centres. SambaNova expects the deal to add significant inference capacity in 2027.
