Executive Briefings Curated
The Infrastructure Race Beneath the Robotics Race
Fei-Fei Li’s world model taxonomy and what it means for automotive and manufacturing
Role pathway
Engineering and platform leaders
Engineering and vehicle or industrial platform leaders choosing architectures, operating systems, safety cases and suppliers for software-defined and autonomous systems.
Curated reading
-
-
Automotive Curated
The First Truck With No Cab Is Running on a German Public Road. The Permit Is the Real Story.
Einride and Lidl just put a driverless Level 4 truck into daily grocery deliveries under the strictest road regulator in Europe. What the KBA accepted matters more than what the truck can do.
-
Automotive Curated
Alpamayo Moves the Cost of Autonomy Out of the Car
NVIDIA’s reasoning model is being read as a chatbot behind the wheel. Its own architecture and its own numbers say the expensive thinking happens before the vehicle ever ships.
-
Artificial Intelligence Curated
NVIDIA’s Open Agent Safety Platform Moves the AI Boundary Out of the Model
OpenShell can restrict what an agent touches. Sentry proposes a second, independent observer in the infrastructure. The harder question is who grants authority when software reaches the factory floor.
-
Physical AI Curated
The Physical AI Stack Is Being Read Backwards
Physical AI is built from intelligence outward. Automotive and manufacturing deploy it from physics inward, where the real bottlenecks emerge.
-
Physical AI Curated
The industry is counting robots. The constraint is actuators, magnets, and a few grams of terbium.
Stop Counting Humanoids. Count Actuators.
Newest first
-
Executive Briefings
Zonal Architecture [Explained]
The shift that lets software run the car is, underneath, a story about copper, fuses and who builds the boxes.
-
Executive Briefings
Software-Defined Vehicle [Explained]
Every automaker uses the phrase. No standard defines it. The architecture behind it decides which carmakers will still control their own products in ten years.
-
Physical AI
The Robot Economics Start at the Accepted Part
RobCo Alfie makes task variability a timely physical AI question. The factory test belongs in exceptions, quality acceptance, and total operating cost.
-
Physical AI
When the Drawing Becomes the Robot Program
FANUC’s AI Welding Agent can write a weld program in the time it takes to photograph a drawing, and nothing in the announcement says who proves the weld is good.
-
Automotive
How QNX Works and Why It Is Difficult to Replicate
The microkernel is visible. The assurance ecosystem is the barrier competitors must recreate.
-
Executive Briefings
NVIDIA Is Building the Common Stack for Physical AI
The 80-entity ecosystem reveals a contest for the infrastructure beneath robot learning and deployment
-
Physical AI
One Video Can Teach the Robot. The Factory Still Has to Trust It.
Physical AI is making robot skills cheaper to create faster than manufacturers can make those skills safe, repeatable, measurable, and certifiable.
-
Physical AI
Atlas Changes the Economics of Training Physical AI
Why World Labs just made real-world observation reusable, how it does it, and why that breakthrough still solves only half of the robotics problem.
-
Automotive
The OTA Update Race Is Growing Up
Automakers are moving from software velocity to controlled fleet operations
-
Physical AI
The Unfiltered Robot Deployment Challenge
Hardware, Not Logic, Is the True Humanoid Bottleneck
-
Automotive
The Real Bottleneck in Vehicle Software Isn’t the Model. It’s the Assembly Line Behind It.
Why the next competitive advantage in automotive and manufacturing AI will belong to whoever can validate and ship the fastest, not whoever trains the smartest model
-
Physical AI
The Architecture Gap Holding Back Multi-Robot Fleets
Collaborative SLAM has been proven in demanding field trials. At fleet scale, it is still far from the default architecture.
-
Physical AI
The Physical AI Stack, Explained
Robotics funding just hit an all time high. Here is what the five layers of Physical AI actually do, and who is building each one.
-
Physical AI
The Goldman map missed the real bottleneck
Humanoid robots are not just an AI race. They are a materials race.
-
Physical AI
The Robot Does Not Suffer. The Robot Pulls Away.
A deep dive on NRE-skin, artificial nociception, and why humanoid robots may need distributed protective reflexes before they can safely work around people.
-
Physical AI
The Humanoid on the Line Has Linux’s Problem
NVIDIA’s robot safety stack splits between Linux and QNX, the same fight SDV compute just had.
-
Automotive
Linux KVM Cannot Anchor an ASIL D Safety Case
A sixteen-year-old kernel bug shows why a cloud hypervisor cannot enforce vehicle safety isolation.
-
Physical AI
Japan’s Quiet Monopoly Inside Every Robot Joint
Two Japanese companies control roughly half the market for the mechanism that lets robots move, and Chinese suppliers are closing in fast.
-
Physical AI
Where Artificial Intelligence Meets Physics
Why reliable motion, not bigger models, may determine who wins the race to Physical AI.
-
Physical AI
Why Robots Still Struggle to Feel
Touch may be the missing sense that turns robots from impressive machines into useful helpers.
-
Industry Events/Shows
InCabin USA 2026 and the Safety System Pointed at You
The industry is learning that the camera inside the car may be as important as the sensors outside it, and far harder to make people comfortable with.
-
Industry Events/Shows
AutoSens USA 2026 Exposed the Unfinished Argument Inside Automated Driving
The industry wants to move from sensors to AI-defined vehicles. Detroit showed why physics, perception, validation, and infrastructure are still refusing to get out of the way.
-
Physical AI
The World Model Race Is Already an Automotive Race
Spatial AI is moving from demo worlds to real roads, robot fleets, and factory floors.
-
Automotive
The Architecture Worked in the Snow. The Standard That Certifies It Is Still Being Written.
RV Tech has validated OTA-capable traction control in Arctic conditions. The harder question is what safety teams can certify before ISO 26262 catches up.
-
Artificial Intelligence
Claude Code’s Real Innovation Wasn’t the Model. It Was the Governance Layer Around It
Anthropic’s coding agent solved a permission and memory problem before a capability problem, and that order matters for automotive software.
-
Executive Briefings
AutoSens USA 2026: The Sensor Stack Strikes Back
Detroit showed that the AI-defined vehicle is not waiting on a bigger model. It is waiting on a cheaper, more robust, more defensible perception stack.
-
Automotive
The Steering Wheel That Has No Wire to the Wheels
Infiniti, Nexteer, and Tesla all decided steer-by-wire was safe enough for production. What they disagree on is where safety actually comes from.
-
Physical AI
The Humanoid Robot Supply Chain Has a Problem Nobody Wants to Price In
Goldman Sachs mapped the robotics ecosystem. The deeper story is not who builds the robots. It is who controls the materials that make them move.
-
Automotive
The Kernel Problem Nobody Wants to Name
QNX Hypervisor is the clean answer when Linux can be treated as an isolated guest. NVIDIA just showed what the harder answer looks like when the safety boundary runs inside Linux itself.
-
Physical AI
The Asset Missing From Every Factory Balance Sheet
Why physical AI shifts competitive advantage toward industrial incumbents, not away from them
-
Physical AI
Physical AI: The 15 Companies Building the Body
Everyone is watching the robot. Almost nobody is watching what makes it move.
-
Automotive
The Software Defined Vehicle Is Already Old
What five industry leaders at IAA 2025 said about the road to AI defined vehicles
-
Automotive
From SDV to ADV
What NVIDIA’s Compute Stack Tells Us About the Next Automotive Architecture Shift
-
Automotive
The Data Moat Wars: Who Actually Owns the Physical AI Flywheel
Seven companies. Five dimensions. One ranking. The answers may surprise you.
-
Automotive
The Incumbents Have the Moat
Why Automotive OEMs Are Positioned for Disproportionate Success in the Physical AI Era
-
Automotive
The Update Paradox
The Software That Was Supposed to Make Cars Better Is Making Them Worse
-
Automotive
Waymo’s Flood Problem Is Not a Software Bug. It Is a Physics Problem.
The world’s most advanced robotaxi fleet halted service across five cities in four days. The reason should concern every AV program on earth.
-
Executive Briefings
The 18 Month Myth
China did not eliminate engineering time. It redistributed engineering pressure.
-
Automotive
The Open Stack That Could Crack Robotaxi
Autoware Foundation is building end-to-end AI for driverless taxis using open-source world models and vision-language-action architectures.
-
Automotive
The Cockpit Is Becoming Someone Else’s Platform
Google’s Gemini rollout is not just an infotainment update. It is a reminder that automakers still have to prove they can own the digital relationship inside the vehicle.
-
Automotive
The Ghost in the Machine When Robotaxis Go Dark
Why a 100 vehicle shutdown in Wuhan may be the most important autonomous vehicle safety signal of the decade
-
Physical AI
The Robot That Clocks In
Humanoid robots are on the factory floor. The question is no longer whether they work. It is who gets to scale them first.
-
