ALCHEMI Toolkit and Toolkit-Ops enable training of AI surrogate models and high-performance atomistic simulations, with a VASP microservice planned to boost GPU throughput.

CUDA-X remains a versatile backbone for AI-driven science, with over 60 updates and broad adoption, including speeding chemistry simulations in chip development through collaborations like TSMC.

The 35 systems are already in mass production, with related Rubin and Vera Rubin deployments underscoring capabilities.

DAQIRI provides a high-performance streaming layer to process real-time detector data, mitigating data loss for faster instruments and has been used by CERN’s ATLAS for real-time AI processing via A-GHOST.

DAQIRI data-streaming library, ALCHEMI microservices, and cuPhoton reference code are designed to democratize real-time GPU-accelerated data processing within CUDA-X ecosystems.

NVIDIA is expanding its AI-hpc footprint in Europe with 35 new systems under development, aimed at empowering millions of researchers and accelerating science, climate research, healthcare, clean-energy decarbonization, and quantum computing.

Leading European centers, including CINECA, Fraunhofer FOKUS, Barcelona Supercomputing Center, and Jülich, are integrating quantum hardware and software via CUDA-Q to enable hybrid quantum-classical workflows.

ALCHEMI accelerates materials discovery with faster simulations and improved memory efficiency, enabling batched molecular dynamics and broader energy and electromagnetics applications.

Among the major deployments, Barcelona Supercomputing Center’s MareNostrum 5 upgrades, BavariaAI’s Blue Swan, IT4LIA AI factory, HLRS’s HammerHAI, and NAISS’s Mimer AI Factory are expected to deliver exaflop-scale training and substantial inference capabilities across a broad European portfolio.

Insider selling has dominated recent activity, with about $224.6 million in shares sold over the last three months and no reported insider purchases, raising investor concerns.

cuPhoton dramatically speeds astronomical data analysis, achieving up to roughly 14,900x faster data loading and up to 8,400x faster signal processing on Rubin Observatory LSST data using NVIDIA Grace Blackwell GPUs.

NVIDIA’s market position and momentum are highlighted alongside potential risks from high valuation and insider selling, balanced against long-term growth from AI infrastructure investments.