In partnership with
Welcome back.
NVIDIA announced partnerships with over a dozen companies on October 28, including Figure, Agility, Foxconn, Toyota, TSMC, Caterpillar, Lucid Motors, and Uber. That covers most of the major players building autonomous factories and robotaxis in America. The first open humanoid foundation model runs on NVIDIA chips. One company now sits at the center of America's physical AI buildout.
October 28 was a big day for physical AI.
NVIDIA Signs Deals with Manufacturers to Build America's Robot Workforce
NVIDIA announced partnerships with Figure, Agility Robotics, Foxconn, Toyota, TSMC, Caterpillar, and Lucid Motors on October 28. The coordinated announcements at NVIDIA's GTC conference in Washington D.C. painted a clear picture: every major player building autonomous factories is using NVIDIA infrastructure.
The subtext: buy American AI infrastructure.
Foxconn is designing its new 242,287-square-foot Houston facility entirely in NVIDIA Omniverse before breaking ground. TSMC is using the same tools for its Phoenix fab. Toyota is simulating its Georgetown, Kentucky facility.
The pattern repeats on the robotics side. Figure's building humanoids with NVIDIA compute for its Helix vision model. Agility's Digit uses NVIDIA Isaac for training. Every humanoid demo at GTC ran on NVIDIA chips.
Here's the question nobody's asking: what happens when one company provides the infrastructure for America's entire manufacturing automation stack? NVIDIA now sits between the robots and the factories, the automakers and the autonomous vehicles, the foundation models and the hardware.
The context everyone cited: $1.2 trillion in announced investments toward U.S. production capacity in 2025. Electronics, pharma, semiconductors. Someone has to staff those factories. NVIDIA's betting it's robots running NVIDIA software on NVIDIA chips.
The reality check: Digital twins and simulations don't build products. Foxconn's Houston facility exists in software. We'll see if it works in hardware. Toyota's Georgetown simulations look good. Actual production lines are messier. The gap between demo and deployment hasn't closed just because the demos got better.
Uber Plans 100,000 Robotaxis Starting 2027
NVIDIA and Uber announced on October 28 that Uber will begin scaling its global autonomous fleet starting in 2027, targeting 100,000 vehicles. The companies will build a joint AI data factory on the NVIDIA Cosmos platform.
Stellantis, Lucid and Mercedes-Benz are collaborating on level 4-ready autonomous vehicles compatible with NVIDIA's DRIVE AGX Hyperion 10 platform. Aurora, Volvo Autonomous Solutions and Waabi are extending the same platform to long-haul freight.
Waymo operates roughly 360,000 trips per week now across five cities. Uber wants 100,000 vehicles operating globally in less than two years. If each vehicle runs 50 trips per week, that's 5 million weekly trips. The scale difference is enormous.
But Waymo operates now. Uber operates in 2027. Two years is a long time in autonomy. Consider what changed between 2023 and 2025.
The strategic difference matters too. Waymo built its own vehicles, sensors, and software stack. Uber is creating a platform for automakers to plug into. Waymo owns the full experience. Uber enables the ecosystem.
Platform plays scale faster if the technology works. Vertical integration controls quality but limits growth. Uber is betting multiple automakers can build vehicles that work with NVIDIA's stack. That's a bigger bet than it sounds.
Why this matters: This isn't Uber building robotaxis. This is Uber building the operating system for robotaxis that automakers build. Think iOS versus iPhone. The question: will automakers accept Uber as the platform layer, or will they want to own the customer relationship directly?
Introducing the first AI-native CRM
Connect your email, and you’ll instantly get a CRM with enriched customer insights and a platform that grows with your business.
With AI at the core, Attio lets you:
Prospect and route leads with research agents
Get real-time insights during customer calls
Build powerful automations for your complex workflows
Join industry leaders like Granola, Taskrabbit, Flatfile and more.
NVIDIA Releases First Open Humanoid Robot Foundation Model
NVIDIA announced Isaac GR00T N1 on October 28, which the company calls the world's first open, fully customizable foundation model for generalized humanoid reasoning and skills.
NVIDIA also introduced Newton, an open-source physics engine under development with Google DeepMind and Disney Research, purpose-built for developing robots.
The "open" designation matters. Most humanoid robot companies guard their AI models. NVIDIA is releasing theirs for anyone to use and customize. This could accelerate development across the industry or fragment it further.
Foundation models for humanoids work differently than language models. They need to understand physics, balance, manipulation, and spatial reasoning. Language models predict the next word. Robot models predict the next motion while keeping the robot upright.
Why this matters: Open models lower the barrier to entry. Smaller companies can build on NVIDIA's foundation instead of starting from scratch. But open also means no moat. NVIDIA is betting on selling chips, not licensing models.
Quick Hits
Locus Robotics hits 6 billion picks at a customer facility in Germany. The warehouse automation company continues expanding in Europe after recording strong growth over the past two quarters. When you're processing 6 billion picks, you've moved past pilot projects.
GlobalData forecasts robotics market will grow from $90.2 billion in 2024 to $205.5 billion by 2030, representing a 15% compound annual growth rate. Exoskeletons will be the fastest-growing segment at 38% CAGR, followed by drones at 19% and logistics robots at 18%.
J&J uses NVIDIA Isaac for surgical robots to simulate how the MONARCH Platform for Urology performs before entering operating rooms. The company is testing device setup and patient interaction in virtual environments using NVIDIA Omniverse and Cosmos platforms.
What I'm Watching
The 2027 robotaxi race: Uber announced 100,000 vehicles by 2027. Waymo targets 1 million weekly trips by 2026. Tesla keeps promising full self-driving. Who actually delivers at scale? Deployment beats promises, but 2027 isn't that far away.
Foundation models for robots: NVIDIA released an open humanoid foundation model. Will this accelerate the industry or create fragmentation? Open models democratize access but remove competitive moats. Every robotics company now needs to decide: build or adopt.
Manufacturing reshoring: $1.2 trillion announced for U.S. production capacity in 2025. Robots are the answer to labor shortages, apparently. But warehouse AMR deployments keep getting delayed. Either the robots aren't ready or the labor shortage isn't real. Which is it?
Forward this to someone building the future of physical AI.
Until next time,
Rob