TeSLA BELIEVES in Humanoid robots as an economic gear change - BUT who will the spoils go to?

Following a discussion with a very smart Telsa engineer, I was recommend to read the above RethinkX article by Tony Seba. This article, which apparently Elon Musk himself recommends, very eloquently argues that humanoid robots will price physical labour so low that disruption becomes inevitable. But even if this is true, and Tesla leads the way, does this make their stock (or any other humanoid robot manufactures) a 'must buy'?

The numbers the article quotes propose entry costs for 'Robotic Labour' that could be under $10 per hour from 2025, on a path to under $1 by roughly 2035 and under 10 cents by 2045. They nicely detail how even a current $200,000 robot unit, working over an expected life of 20,000 hours, lands at $10 per hour cost of labour today - cheaper than most human labour already. With machine's expected to get cheaper, and more competent, in the future this gap should only compound they argue. While these claims are likely a victim of "over forecasting impact in the short term, and under forecasting impact in the long term", they are certainly interesting (and grounded) enough to make it worth being thoughtful on...

Overall the article rests on the observation of the current cost stack falling - across sensors, compute and AI control - with RethinkX's in-house estimates placing early operating cost between $2 and $10 per hour as manufacturing scales. They argue that if history of 10x cost collapses holds, and adoption follows the typical S-curve, we will see incumbent human tasks out-competed in many settings long before full human parity. However, this is a system level claim, not necessarily factory-floor impact. Task mix, safety regimes, liability and integration bottlenecks will obviously vary by sector. Further unit economics depend on duty cycle - with any robot idle half the day likely not living up to its "$10 per hour" potential, with obsolescence likely kicking in prior to the 20,000 hours 'rated life'.

But what I am most interested in is where will the benefits of the above, assuming it does come true, accrue to? In my view even if humanoid robots drive labour costs toward zero, that does not guarantee excess returns for robot makers. It more likely most value will shift to the firms that deploy them at scale, i.e. the productive users.

As volumes rise, designs converge, components standardise, and the bill of materials becomes a shopping list. Margins migrate from boxes to integration and service. Buyers gain bargaining power because they purchase fleets, dual-source parts, and negotiate based on economic generation capacity (where they have the benefit of asymmetric information). Switching costs between robot producers may also be lower than they first seem: task definition and policies live in software and data, with an increasing amount of the control systems now also being cross hardware (such as ROS, Isaac Sim, etc). In manufacturing terms, the robot arm is a fixture. The process is the moat. However, deployment is where systems engineering compounds. The winner is the company that rewrites workflows, tunes duty cycles, and couples robots to data, QA and scheduling. That firm converts cheap capability into throughput, reliability and inventory turns. It benefits from learning curves specific to its products and sites. Those gains are hard to copy because they are entangled with layout, SKU mix and proprietary know-how.

So overall, I see investing in robot companies equities is a one-way bet. If the technology under-delivers, the humanoid robot designers and manufacturers loose out. But, if it succeeds, competitive entry, open standards and buyer power cap their returns. By contrast, investing in the adopters is a 'win-not-lose' proposition. If the robotic promise flops, the business remains unchanged. If robots work though, operating leverage improves and cash costs fall - all without you having had to invest in the dream upfront.