Top AI GPU Sessions at NVIDIA GTC 2026
NVIDIA GTC 2026 is right around the corner ~ March 16–19, and if you can’t make it to San Jose in person, the good news is that some of the most impactful sessions are available virtually. From Jensen Huang’s keynote (where he’s teased unveiling chips the “world has never seen before”) to deep dives on agentic AI, robotics, sovereign AI compliance, and manufacturing intelligence – virtual attendees get access to world-class content without leaving their desk.
I’ve gone through the GTC 2026 Session Catalog and handpicked the virtual sessions that matter most for developers building at the intersection of AI, accelerated computing, and modern infrastructure. Here’s your guide to making the most of GTC 2026 from anywhere.
Jensen Huang’s Keynote: The Must-Watch Session
GTC 2026 Keynote [S81595]
📅 Monday, March 16 | 11:00 AM – 1:00 PM PT | Free Livestream | Keynote Page
Every GTC starts with Jensen’s keynote, and this year it’s going to be massive. In a recent interview with Korean media, Jensen revealed that NVIDIA has prepared “several new chips the world has never seen before,” with at least one chip that will “surprise the world.” With Vera Rubin already in full production since CES 2026 featuring Rubin GPUs delivering 5x inference performance over Blackwell with 288GB of HBM4 memory — the keynote could unveil everything from next-gen Rubin derivatives to entirely new architectures.
For developers, the keynote typically sets the direction for the entire NVIDIA ecosystem — new SDKs, platform updates, developer tooling, and the roadmap for accelerated computing. Don’t miss the GTC Live 2026 Pregame show starting at 8:00 AM PT featuring special guests and pre-keynote discussions.
💡 Virtual attendees: Just visit nvidia.com/gtc/keynote on March 16 to watch the livestream for free.
Agentic AI & Generative AI Sessions
Agentic AI is the hottest theme in the developer world right now, and GTC 2026 has several virtual sessions spanning code generation, RAG, 3D models, and real-world agent deployments.
Agentic Systems for End-to-End Surgical Workflows [S81941]
Track: Agentic AI / Generative AI – Speech Generation
AI agents are making their way into some of the most high-stakes environments imaginable – surgical operating rooms. This session explores how agentic systems are being built to support end-to-end surgical workflows, from pre-operative planning through real-time intraoperative assistance. It’s a powerful look at how multi-step reasoning agents are being applied where precision and safety are non-negotiable.
Why it matters: For developers interested in building agents for mission-critical domains – healthcare, robotics, industrial safety — this session demonstrates what production-grade agentic AI looks like when lives are on the line.
Teach AI to Code in Every Language With NVIDIA NeMo [S82306]
Track: Agentic AI / Generative AI – Code / Software Generation
NVIDIA NeMo is one of the most powerful frameworks for building and customizing large language models, and this session focuses specifically on training AI to generate code across multiple programming languages. If you’re working on code generation, developer tooling, or building coding assistants, this session walks through the NeMo-based approach to multilingual code intelligence.
Why it matters: As AI coding assistants become mainstream, understanding how to train and fine-tune models for code generation across diverse languages is a critical capability. NeMo gives you the framework to do it at scale.
Supercharging Responsible AI Innovation: Compute, Synthetic Data, and Regulatory-Informed Design [S82044]
Track: Agentic AI / Generative AI – Code / Software Generation
Building responsible AI isn’t just about guardrails – it’s about designing systems from the ground up with compliance, synthetic data strategies, and compute-efficient architectures. This session covers how to balance innovation velocity with responsible AI principles, including regulatory-informed design patterns that are becoming essential as AI regulations tighten globally.
Why it matters: Every developer shipping AI in production needs to think about responsibility and compliance. This session gives you a practical framework for doing it without slowing down innovation.
Agentic AI for the Next Generation of Financial Advisors [S81718]
Track: Agentic AI / Generative AI – Retrieval-Augmented Generation (RAG)
The financial services industry is one of the most promising arenas for agentic AI, and this session explores how RAG-powered agents are being built to support the next generation of financial advisors. From real-time market intelligence to personalized client recommendations, the session covers architectures and patterns for building financial AI agents that are accurate, compliant, and trustworthy.
Why it matters: Great session for anyone building RAG-based agents in regulated industries. The patterns here apply well beyond finance – to legal, healthcare, and any domain where accuracy and auditability are critical.
Manufacturing Intelligence: Agentic AI to Unlock the Next Level of Performance [S82204]
Track: Agentic AI / Generative AI – 3D Model Generation
Manufacturing is undergoing a massive transformation with AI, and this session shows how agentic AI is being combined with 3D model generation to unlock new levels of manufacturing performance. Think intelligent agents that can reason about physical processes, optimize production lines, and generate 3D models for simulation and validation.
Why it matters: For developers working at the intersection of AI and physical systems — manufacturing, supply chain, industrial IoT – this session bridges the gap between digital intelligence and physical-world outcomes.
Watch Parties
GTC 2026 features curated Watch Party sessions that are perfect for virtual attendees — designed to be community viewing experiences with discussion and context.
Watch Party: Nemotron Unpacked – Build, Fine-Tune, and Deploy NVIDIA’s Open Models [WP81719]
Track: Agentic AI / Generative AI – Text Generation
NVIDIA’s Nemotron family of open models has been gaining serious traction in the developer community. This watch party unpacks everything you need to know about building with, fine-tuning, and deploying Nemotron models from architecture details to practical deployment patterns.
Why it matters: Open models are the foundation of customizable AI. If you’re building applications on top of open-weight LLMs, understanding NVIDIA’s Nemotron ecosystem gives you a powerful alternative to closed-source models.
Watch Party: NVIDIA 101 – From Chips to Software: How NVIDIA’s AI Platform Enables the AI Revolution [WP81708]
Track: Agentic AI / Generative AI – Text Generation
A great foundational session for developers who want to understand the full NVIDIA AI stack – from silicon to software. This watch party covers how NVIDIA’s hardware and software platforms work together to enable the AI revolution, making it an excellent starting point for anyone new to the NVIDIA ecosystem or looking to deepen their understanding of the end-to-end platform.
Why it matters: Understanding the full stack – chips, CUDA, libraries, frameworks, and deployment – gives you a significant advantage when optimizing AI workloads. This session connects all the dots.
Trustworthy AI & Compliance
AI in the Age of Regulation: Build Trust Through Compliance [S81627]
Track: Trustworthy AI / Cybersecurity – Security for AI
As AI regulations like the EU AI Act take effect and new frameworks emerge globally, compliance is no longer a “nice to have” – it’s a requirement. This session covers how to build AI systems that meet regulatory requirements while maintaining trust with users and stakeholders. From documentation and auditability to risk assessment frameworks, it’s a practical guide to compliance-first AI development.
Why it matters: Critical for any team deploying AI in production, especially in regulated markets. As AI governance frameworks tighten worldwide, understanding how to build compliant systems from day one saves you from costly retrofits later.
Robotics & Physical AI
Build Physical Intelligence: Open Collaboration Across Robotics Ecosystems [S81509]
Track: Robotics – Humanoid Robots
Physical AI is one of NVIDIA’s biggest bets, and this session explores how open collaboration across robotics ecosystems is accelerating the development of physical intelligence. With NVIDIA’s Isaac platform, Cosmos for synthetic training data, Isaac GR00T for humanoid robot foundation models, and the Newton physics engine – the tools for building physical AI are becoming increasingly accessible to the broader developer community.
Why it matters: The convergence of AI and robotics is creating entirely new categories of applications. This session shows how open ecosystems and shared infrastructure are making physical AI development practical for more teams, not just large robotics labs.
Simulation & Digital Twins
Customer-Centric Innovation in Automotive [S82116]
Track: Simulation / Modeling / Design – Industrial Digitalization / Digital Twin
The automotive industry is leveraging digital twins and simulation to drive customer-centric innovation – from personalized vehicle design to predictive maintenance and connected car experiences. This session explores how industrial digitalization and digital twin technologies are reshaping how automotive companies innovate, test, and deliver products.
Why it matters: For developers working in automotive, manufacturing, or industrial simulation, this session shows how digital twin technology is moving from experimental to production-grade, enabling faster iteration and better customer outcomes.
My Top 5 Virtual Picks for Developers
If you’re short on time, here are the five virtual sessions I’d prioritize:
| # | Session | Why |
| 1 | GTC 2026 Keynote [S81595] | Non-negotiable. Free livestream. Jensen’s keynote sets the stage for everything — expect new chip reveals. |
| 2 | Teach AI to Code in Every Language With NeMo [S82306] | Essential for anyone building AI coding tools or working with code generation at scale. |
| 3 | Agentic AI for Financial Advisors [S81718] | Best-in-class example of RAG-based agents in a regulated industry — patterns apply everywhere. |
| 4 | AI in the Age of Regulation [S81627] | Compliance is now mandatory. This session teaches you how to build trust-first AI systems. |
| 5 | Watch Party: Nemotron Unpacked [WP81719] | Open models are the future. Learn to build, fine-tune, and deploy NVIDIA’s Nemotron models. |
Quick Reference: All Virtual Sessions at a Glance
| Session ID | Session Title | Track |
| S81595 | GTC 2026 Keynote | Keynote |
| S81941 | Agentic Systems for End-to-End Surgical Workflows | Agentic AI — Speech Generation |
| S81509 | Build Physical Intelligence: Open Collaboration Across Robotics Ecosystems | Robotics — Humanoid Robots |
| S82306 | Teach AI to Code in Every Language With NVIDIA NeMo | Agentic AI — Code / Software Generation |
| S82044 | Supercharging Responsible AI Innovation | Agentic AI — Code / Software Generation |
| S81627 | AI in the Age of Regulation: Build Trust Through Compliance | Trustworthy AI / Cybersecurity |
| S81718 | Agentic AI for the Next Generation of Financial Advisors | Agentic AI — RAG |
| S82116 | Customer-Centric Innovation in Automotive | Simulation — Digital Twin |
| S82204 | Manufacturing Intelligence: Agentic AI | Agentic AI — 3D Model Generation |
| WP81719 | Watch Party: Nemotron Unpacked | Agentic AI — Text Generation |
| WP81708 | Watch Party: NVIDIA 101 — From Chips to Software | Agentic AI — Text Generation |
🔍 Search for any session ID in the GTC 2026 Session Catalog for full details, schedules, and speaker information.
How to Attend Virtually
NVIDIA GTC 2026 runs March 16–19, 2026. The virtual experience is free and includes:
- ✅ Jensen Huang’s Keynote Livestream – Free, no registration required
- ✅ Select Sessions, Talks & Tutorials – Available on-demand after the event
- ✅ Live Q&A – Available for select virtual sessions
- ✅ Watch Party Sessions – Community viewing experiences with curated content
To get started:
- Visit nvidia.com/gtc and register for a free virtual pass
- Browse the Session Catalog and add sessions to your schedule
- Watch the keynote live on March 16 at nvidia.com/gtc/keynote
Whether you’re building AI agents, exploring physical AI, navigating AI compliance, or diving into open models – GTC 2026’s virtual sessions have you covered. The best part? It’s free.
