Was this email forwarded to you? Sign up here for Telemetry.

Modern cars are starting to operate as robot platforms, and that foreshadows how such systems architecture will spread to every industry that moves physical things. A Boston Dynamics Atlas humanoid robot and a Tesla Cybertruck are closer architectural relatives than the Cybertruck is to any vehicle built before 2020. (Our upcoming Robotics Core covers this architecture across over a dozen interactive modules, spanning actuator physics to perception stacks.)

Yesterday, Tesla confirmed that Model S and Model X production is over. Roughly 600 vehicles remain in inventory worldwide. The Fremont factory lines that built those cars are converting to manufacture Optimus humanoid robots: one million units per year at $20,000 each, with public sales beginning in 2027. The company also reportedly placed a $685 million actuator order with Chinese supplier Sanhua Intelligent Controls, enough components for roughly 180,000 robots. A dedicated Optimus facility is breaking ground at Giga Texas, targeting 10 million annual units. The company plans to unveil Optimus v3 this quarter, its first design meant for mass production. While Tesla is literally converting automotive facilities to robot-production facilities, we’ve been seeing evidence from the cars themselves that a change was coming; the internal wiring tells a backstory.

Back in 2022, Ford CEO Jim Farley described what his engineers found when they disassembled a Tesla Model Y: "We had prejudice,“ but the proof was sprawled in front of them: the Mach-E wiring harness was 70 pounds heavier and 1.6 kilometers longer than Tesla's equivalent. (For mass-production, these differences meaningfully compound.)

The Cybtertruck’s distinctive design widened the gap further: 155 wires versus a traditional vehicle's 400-500, achieved through 48-volt architecture, Ethernet replacing the CAN (Controller Area Network) bus, and zone controllers handling local devices rather than routing everything to a central fuse box. (Munro’s 🎥 teardown series is fascinating.) S&P Global estimates Tesla holds a five-year lead in electrical/electronic architecture over every other automaker. Centralized compute, zonal controllers, software-defined actuation, sensor fusion through a unified data bus: these are not traditional automotive concepts but rather the foundational architecture of robotic systems, and they apply whether the robot carries passengers, welds chassis, or walks on two legs.

Ford is (now) converging on the same architecture from the other direction. The company's next-generation Universal Electric Vehicle platform adopts 48V zonal architecture with five in-house zone controllers and a wiring harness 1.2 kilometers (4,000 feet!) shorter than the Mach-E's. The first vehicle of this type, a $30,000 electric pickup, arrives in 2027.

The tier-1 automotive suppliers are following the architecture, too. At CES 2026 (the Consumer Electronics Show), Hyundai Mobis (developer of brake systems, chassis modules, and headlamps for Hyundai and Kia) announced it will supply actuators for the new Boston Dynamics Atlas. Actuators represent more than 60% of a humanoid robot's material cost, and have been a dogged limiter of robotics for quite a while. (See our Thrust on RISE Robotics for an interview with the founders where they describe this challenge and how they approached it.)

Boston Dynamics has since asked Mobis for five additional core components: grippers, sensors, controllers, and battery packs. Zack Jackowski, GM of Atlas, said this "allows us to access the well-established cost structures and scale potential of the automotive industry."

Mobis is considering a U.S. factory for robot parts. All 2026 Atlas allocations are already committed, with fleets shipping to Hyundai's Robotics Metaplant and Google DeepMind. The group plans to produce 30,000 robot units per year by 2028. Last month, Schaeffler (precision bearings, gears, and motors for automakers worldwide) partnered with Leju Robotics in China and unveiled an all-in-one actuator for humanoids, establishing a dedicated embodied intelligence unit in Suzhou.

The appetite driving these moves extends well beyond humanoid robots. Automotive, construction, logistics, defense, agriculture: every one of these industries needs greater digitization, greater precision of control, and eventually greater autonomy in how equipment moves and operates. The system architecture that delivers those capabilities is the same architecture regardless of form factor.

A Goldman Sachs survey of nine Chinese supply chain companies found what the bank called a "capacity-first" strategy: suppliers like Sanhua, Tuopu Group, and BYD are building annual production capacity for 100,000 to one million robot-equivalent units, ahead of firm orders. Global humanoid shipments are forecast to breach 50,000 in 2026, a 700% year-over-year surge. Chinese manufacturers control roughly 70% of the global humanoid component supply chain. This has a lot of implications for non-Chinese players, particularly the United States and a push for manufacturing ‘re-industrialization’. (I’ll certainly cover aspects of this in a future edition because there’s been a ton of work in this domain, mostly behind-the-scenes or at least poorly covered by the media.)

The robot system architecture works well at human scale, where electric motors and ball-screw actuators handle the loads. At multi-ton scale, a bottleneck remains. Excavators, cranes, forklifts, and munitions handlers are still locked in hydraulic systems: pressurized toxic fluid pumped through hoses, usually running on diesel powerplants. Hundreds of millions of liters of hydraulic oil leak into the environment every year.

Conventional electric actuators (ball-screw, lead-screw) hit practical force limits at a few tons, which is why the heavy equipment sitting on every construction site and forward operating base still runs on fluid power.

Belt-driven and other electromechanical architectures are bypassing those limits, so there will be more investment and deployment opportunities as we get robots prepped for a new (heavy)weight class! RISE Robotics holds the Guinness World Record for Strongest Robotic Arm Prototype at 7,015 pounds using Beltdraulic actuators, nearly 2,000 pounds beyond the previous benchmark held by Fanuc. Their actuators achieve 70-85% efficiency versus 21% for hydraulic systems (regenerative power capture FTW!), while eliminating the entire hydraulic balance-of-plant: the pump, reservoir, hoses, and fluid. Our Beyond Hydraulics deep-dive examined the technology and the investment thesis in detail.

The capability difference between a ‘robot-ified’ platform and, well, traditional platforms extends beyond raw efficiency. An electrified actuator with integrated sensors and a digital bus can report its own wear state, feed data to predictive maintenance models, recover energy during load lowering, and accept software updates that change its force profile. A hydraulic cylinder tells you it's failing only after it springs a leak.

Electrified actuation also brings the digitization, the control, and the maintainability that makes the whole robot system architecture work at industrial scale. The $40-50 billion hydraulic equipment market is the addressable opportunity, and it's available pretty much now.

Back to the humanoid element: Famed MIT roboticist Rodney Brooks argues the humanoid boom is a bubble, and the criticism has merit on adoption timelines. But the architecture thesis and the actuation thesis don't depend on humanoids. The same components, the same system design, the same supply chain serve industrial arms, construction equipment, defense platforms, and autonomous vehicles whether humanoids ship at volume in 2027 or 2032.

So, if you're a component supplier selling to automotive manufacturers: the addressable market just expanded by an order of magnitude. An actuator architecture that works in a truck liftgate, a construction crane, and a robotic arm has a fundamentally different venture trajectory than one designed for a single end market.

If you’re an LP or portfolio strategist: In the iPhone supply chain, Apple and, in many markets, the carriers captured the single biggest slices of value, but a select group of component suppliers with defensible IP in the bill of materials (e.g., Qualcomm, Corning, TI, TSMC) also captured outsized, durable economics relative to assemblers like Foxconn. Mobis is positioning as the Foxconn of humanoid robots. The question for your portfolio: who becomes the Corning, the Qualcomm, the TI of this stack?

Google absorbed Intrinsic, its robotics software subsidiary, and announced a Foxconn joint venture to deploy robots in factories; Sundar Pichai called the platform "Android for robotics" • Boston Dynamics and Google DeepMind formed a partnership to build Gemini Robotics foundation models for Atlas, locking the ‘reasoning layer’ to specific hardware. • Mobileye, Intel's autonomous driving subsidiary, acquired Mentee Robotics for $900 million, with a focus on repositioning vehicle perception for humanoid robots • EngineAI launched the T800 humanoid at $25,000 with NVIDIA Jetson Thor in the higher-spec editions, shipping June 2026.

No One Builds Alone.

/N1BA

Telemetry is written by JMill of The End Effector.