Jensen Huang: NVIDIA GTC Keynote, Mar 16, 2026

At the keynote on Mar 16, 2026, Jensen Huang said 29 of the 35 words Kalshi and Polymarket listed; Parrot’s count agreed with all 35 settlements. Quotes below are at most 25 words each.

Updated

Markets

35 words · 29 settled yes · our count agreed on 35 of 35

backtest The probabilities on settled rows are a walk-forward backtest: the model as of before each event, computed after Parrot went live; they were not published at the time.

Listed words: result, our count under the market rule, the count over every speaker, whether they agree, and the probability (a walk-forward backtest on settled rows)
WordResultOur countAll speakersAgreesProbabilityMarket
AI / Artificial Intelligence 10+ timesYes178178✓–n=1 too little history for a probability–
Agent / Agentic Yes3737✓–new new word, no probability yet–
Million / Billion 3+ timesYes2929✓–new new word, no probability yet–
Vera Rubin Yes2828✓–new new word, no probability yet–
GPU 5+ timesYes2424✓–new new word, no probability yet–
Groq / Grok Yes2222✓58%n=2 from settlement history–
Data Center Yes2222✓–n=1 too little history for a probability–
OpenClaw Yes2121✓–new new word, no probability yet–
Partner Yes1616✓–new new word, no probability yet–
Blackwell Yes1313✓85%n=3 from settlement history–
Training Yes1212✓–new new word, no probability yet–
Open Source Yes1010✓82%n=2 from settlement history–
Trillion Yes1010✓–n=1 too little history for a probability–
Google Yes1010✓–new new word, no probability yet–
Invest / Investment Yes88✓–n=1 too little history for a probability–
OpenAI Yes88✓–new new word, no probability yet–
Cosmos Yes77✓–new new word, no probability yet–
Deep Learning Yes77✓–new new word, no probability yet–
Anthropic Yes55✓–n=1 too little history for a probability–
Autonomous Yes55✓–new new word, no probability yet–
Quantum Yes44✓60%n=3 from settlement history–
Self Driving Yes44✓53%n=3 from settlement history–
Alpamayo Yes33✓–n=1 too little history for a probability–
Humanoid Yes33✓53%n=3 from settlement history–
Dynamo Yes33✓–new new word, no probability yet–
Microsoft Yes33✓–new new word, no probability yet–
Hyperscaler Yes33✓10%n=3 from settlement history–
Meta Yes22✓–new new word, no probability yet–
Photon / Photonics Yes11✓–new new word, no probability yet–
Elon / Musk No00✓55%n=4 from settlement history–
Super Bowl No00✓–new new word, no probability yet–
FSD / Full Self-Driving No00✓–new new word, no probability yet–
Trump No00✓28%n=3 from settlement history–
Stargate No00✓13%n=2 from settlement history–
Domestic No00✓10%n=3 from settlement history–

Market prices are mid quotes from Kalshi and Polymarket, shown for context, may be delayed by several minutes and are not a signal. Quotes as of Oct 21, 2026 04:22 UTC.

Transcript

quotes of at most 25 words

Captions · machine-attributed · NVIDIA (YouTube) · Source · Quotes of at most 25 words; the full recording is on the publisher's official channel.

Machine-attributed. Speaker turns are labelled by a model, not the publisher.

12 matches for “Training”; 5 quoted, at least 25 words apart.

Jensen Huangspeaker51:09

…It has to generate tokens. It has to inference. It's way past training. Now it's in the in the field of inference. So the the…

Jensen Huangspeaker54:28

…We wanted to make sure that not only were we good at training and post training that we were incredibly good at every single phase…

Jensen Huangspeaker1:00:09

…performance and energy efficiency. We've also been able to use NVFP4 for training. So NVLink 72 NVFP4 the invention of Dynamo TensorRT-LLM, a whole bunch…

Jensen Huangspeaker1:09:54

…runs over Spectrum for Ethernet with three scaling laws in full steam pre-training, post training and inference, and now Agentic systems compute, demand continues to…

Jensen Huangspeaker1:58:01

…We build and release six families of open frontier models, plus the training data, recipes, and frameworks to help developers customize and adopt new leaderboard…

Where the listed words were said

Jensen Huangspeaker“AI / Artificial Intelligence (10+)” · 0:25

…A new kind of factory generator of tokens. The building blocks of AI. Tokens have opened a new frontier, turning data into knowledge and drawing…

Jensen Huangspeaker“Agent / Agentic” · 18:02

…be used by AI, and AI is going to be much, much faster than us. Future agents are going to use structured databases as well.

Jensen Huangspeaker“Million / Billion (3+)” · 7:23

…has taken us 20 years to now have built up hundreds of millions of GPUs and computing systems around the world that run CUDA. We…

Jensen Huangspeaker“Vera Rubin” · 1:10:57

…an incredible new addition, the Groq 3 LPX rack, tightly connected to Vera Rubin Groqs LPU's massive on chip SRAM, a token accelerator to the…

Jensen Huangspeaker“GPU (5+)” · 8:58

…so the application reach is so great that once you install NVIDIA GPUs, the useful life of it is incredibly high. It is also one…

Jensen Huangspeaker“Groq / Grok” · 1:15:08

team. NVLink completely cooled. This is the brand new Groq system and I'll show you a little bit more about it. This system eight Groq…

Jensen Huangspeaker“Data Center” · 31:33

…algorithm in whatever scenario it wants to be deployed, whether it's a data center, cloud on prem at the edge, or in a robotic system.

Jensen Huangspeaker“OpenClaw” · 1:46:40

…and, um, he, he wrote a piece of software, uh, it's called OpenClaw. And and, um, I don't know if he realized, uh, how successful…

Jensen Huangspeaker“Partner” · 26:30

NVIDIA's integrated really deeply into AWS. They were our first cloud partner. ________________ Microsoft Azure. NVIDIA's A100 supercomputer was the first one we built was…

Jensen Huangspeaker“Blackwell” · 52:49

…saw $500 billion of very high confidence demand and purchase orders for Blackwell and Rubin through 2026. I said that last year. Now, I don't…

Jensen Huangspeaker“Training” · 2:06:08

We have three computers the training computer, the synthetic data generation and simulation computer, and of course, the robotics computer that sits inside the robot…

Jensen Huangspeaker“Open Source” · 6:51

…intelligence today. Thousands of tools and compilers and frameworks and libraries in open source. There's a couple of hundred thousand public projects. CUDA literally is…

Agents: this page as TSV at /api/v1/appearances/jensen-huang:2026-03-16:keynote · words: /api/v1/brief?appearance=jensen-huang:2026-03-16:keynote · columnar JSON in #parrot-data · llms.txt · MCP