Jiageng Mao , University of Southern California — Solving complex physical AI problems by using diverse priors from internet-scale data to enable robust, generalizable intelligence for embodied agents in the real world.
Liwen Wu , University of California San Diego — Enriching realism and efficiency in physically based rendering with neural materials and neural rendering.
Manya Bansal , Massachusetts Institute of Technology — Designing programming languages for modern accelerators that enable developers to write modular, reusable code without sacrificing the low-level control required for peak performance.
Sizhe Chen , University of California, Berkeley — Securing AI in real-world applications, currently securing AI agents against prompt injection attacks with general and practical defenses that preserve the agent’s utility.
Yunfan Jiang , Stanford University — Developing scalable approaches to build generalist robots for everyday tasks through hybrid data sources spanning real-world whole-body manipulation, large-scale simulation and internet-scale multimodal supervision.
Yijia Shao , Stanford University — Researching human-agent collaboration by developing AI agents that can communicate and coordinate with humans during task execution, and designing new human-agent interaction interfaces.
Shangbin Feng , University of Washington — Advancing model collaboration: multiple machine learning models, trained on different data and by different people, collaborate, compose and complement each other for an open, decentralized and collaborative AI future.
Shvetank Prakash , Harvard University — Advancing hardware architecture and systems design with AI agents built on new algorithms, curated datasets and agent-first infrastructure.
Irene Wang , Georgia Institute of Technology — Developing a holistic codesign framework that integrates accelerator architecture, network topology and runtime scheduling to enable energy-efficient and sustainable AI training at scale.
Chen Geng , Stanford University — Modeling 4D physical worlds with scalable data-driven algorithms and physics-inspired principles, advancing physically grounded 3D and 4D world models for robotics and scientific applications.
We also acknowledge the 2026-2027 fellowship finalists:
Key considerations
- Investor positioning can change fast
- Volatility remains possible near catalysts
- Macro rates and liquidity can dominate flows
Reference reading
- https://blogs.nvidia.com/blog/graduate-fellowship-recipients-2026-2027/#content
- https://www.nvidia.com/en-us/
- https://blogs.nvidia.com/?s=
- Linux's contemporary filesystem mount API went without documentation for six years — latest man-page package finally adds content for 2019 code
- Chinese memory maker CXMT prepares $4.2 billion USD IPO to take advantage of tight memory market — company lays out path to profitability as DRAM demand skyrock
- Nvidia's CUDA Tile examined: AI giant releases programming style for Rubin, Feynman, and beyond — tensor-native execution model lays the foundation for Blackwel
- Grab this 4K gaming PC with an RTX 5080 and 9800X3D for just $2,499 before prices skyrocket — $200 saving on powerful Skytech rig with 32GB of DDR5 RAM and 1TB
- 3 Ways NVIDIA Is Powering the Industrial Revolution
Informational only. No financial advice. Do your own research.