engblogs

OpenAI and NVIDIA form a landmark partnership to deploy 10 gigawatts of NVIDIA AI systems, supported by up to $100 billion investment, advancing next-generation AI infrastructure and computing power.

SchoolAI leverages advanced OpenAI models like GPT‑4.1 to create a safe, observable, and personalized AI platform that supports teachers and students across 1 million classrooms worldwide.

Exploring the latest advancements and safety approaches in American-developed AI technologies and platforms.

CNA is pioneering an AI-driven transformation in its newsroom by integrating advanced AI tools like ChatGPT to enhance journalistic workflows, improve public service journalism, and foster widespread adoption and training across its teams.

MIT entrepreneurs integrate AI tools like Jetpack and custom bots to accelerate startup development while emphasizing the enduring importance of customer validation and fundamental entrepreneurial principles.

MIT researchers developed SCIGEN, a tool that guides generative AI diffusion models to design quantum materials with specific geometric constraints, accelerating discovery of breakthrough materials for quantum computing and other advanced technologies.

MIT researchers receive AI for Math grants to integrate the LMFDB database with the Lean4 mathlib library, enhancing automated theorem proving and accelerating mathematical discovery through AI and formal verification.

Ink & Switch joins the ARIA Safeguarded AI Programme to develop collaborative AI-human modeling interfaces, welcomes new researchers, and shares innovative explorations on Programmable Ink technology.

Modular Platform 25.6 unifies GPU programming across NVIDIA, AMD, and Apple Silicon with enhanced performance, developer tools, and a single compute layer for AI workloads.

The future of generative AI lies in advancements like world models, ethical guardrails, and its integration into robotics and various industries to enhance human efficiency and societal benefit.

Exploring practical implementation strategies, optimizations, and scaling techniques for leveraging JAX on GPUs to enhance enterprise machine learning workflows with high performance and portability.

Exploring advanced optimizations including Cluster Launch Control and persistent kernel techniques to achieve state-of-the-art matrix multiplication performance on NVIDIA Blackwell GPUs, surpassing existing benchmarks with 1772 TFLOPs.