Explore how the latest technologies are shaping our world, from groundbreaking discoveries to transformative sustainability efforts. The NVIDIA AI Podcast shines a light on the stories and solutions behind the most innovative changes, helping to inspire and educate listeners. More information: https://ai-podcast.nvidia.com/
What this show covers
A/B testingData infrastructureGPU accelerationApache SparkGoogle CloudEnterprise AIOpen-weight modelsAI agent permissionsMixture-of-expertsModel customizationopen source roboticsJetsonrobot learningembodied AIIsaac Sim
What listeners will find here
How Snap turned idle GPUs into a 76%-cheaper Spark engine
Snap’s data team explains how it processes more than 10 petabytes of experimentation data every morning using GPU-accelerated Apache Spark on Google Cloud. The episode focuses on A/B testing at massive scale, reusing overnight inference GPUs, and the infrastructure changes that cut costs, memory use, and disk spill dramatically.
Mistral’s enterprise AI play: open weights, faster training, and safer agents
Open-weight AI models, enterprise customization, and agent permissions are at the center of this conversation with Mistral CTO Timothée Lacroix. The episode explores why Mistral sees openness as a competitive advantage, how Forge targets production use cases, and what faster GB200 training could unlock.
How open-source robot arms are making embodied AI practical and affordable
Open-source robotics takes center stage as Seeed Studio explains how low-cost Jetson-powered arms, OpenClaw, and NVIDIA Isaac Sim can turn embodied AI into something makers, students, and small businesses can actually use. The conversation focuses on accessibility, teachable robot skills, and responsible real-world deployment.
Why cost per token may matter more than FLOPS in the AI business race
AI tokenomics is the focus here, with Shruti Koparkar explaining how businesses can connect infrastructure choices to revenue. The episode lays out a four-part framework for token utility, supply, demand, and monetization, then ties it to NVIDIA Blackwell’s efficiency and lower token costs.
How Instacart wants to turn the grocery cart into an AI shopping assistant
Instacart’s smart cart strategy is the focus here, from Caper Cart’s real-time item recognition to the broader push to digitize physical stores. David McIntosh explains how edge AI, personalized recommendations, and store-ops automation could reshape grocery shopping over the next decade.
Episodes
More like NVIDIA AI Podcast
The a16z Show
Andreessen Horowitz
The AI Daily Brief: Artificial Intelligence News and Analysis
Nathaniel Whittemore
Bloomberg Tech
Bloomberg
The Vergecast
The Verge
All-In with Chamath, Jason, Sacks & Friedberg
All-In Podcast, LLC
Sourcery
Sourcery with Molly O'Shea
The Other Hand
Jim Power & Chris Johns
Decoder with Nilay Patel
The Verge
Latent Space: The AI Engineer Podcast
Latent.Space
Pivot
New York Magazine