NVIDIA AI Updates: May 10, 2026

1. NVIDIA releases open-source quantum AI model family Ising

NVIDIA. Jensen Huang announced Ising, billed as the world’s first family of open-source quantum AI models, positioning NVIDIA in the emerging quantum computing stack alongside its classical AI portfolio. The release reinforces NVIDIA’s strategy of providing software tooling for hybrid classical/quantum workloads even as commercial quantum hardware remains nascent. Source

2. NVIDIA tops $40B in AI equity bets with IREN and Corning deals

NVIDIA. NVIDIA disclosed an agreement to invest up to $2.1 billion in data-center operator IREN, one day after committing up to $3.2 billion in glassmaker Corning. The Corning pact funds three new US facilities producing optical components for NVIDIA rack-scale systems, signaling a broader shift from copper to fiber interconnects in next-gen AI clusters. Source

3. AI Red Team uses grammar-constrained decoding to improve Bash generation

NVIDIA. A new developer-blog post from NVIDIA’s AI Red Team demonstrates how grammar-constrained decoding helps small language models produce syntactically valid Bash commands for agentic use cases. The technique narrows the decoder’s token space to the shell grammar, reducing malformed pipelines and unsafe constructs without requiring fine-tuning. Source