New Humanoid Robot Solution: Hydraulic Muscles Cost One-Tenth of Motor-Driven Systems
Clone Robotics replaces motor-driven cables with hydraulic artificial muscles—pressurized elastic tubes inside braided sleeves that contract directly. The CEO claims this costs one-tenth as much as motor-driven humanoid robots.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Anthropic commits 5% equity to public benefit funds
Anthropic announced it will invest 5% equity into a public benefit fund valued at approximately $100 billion at the company's $2 trillion valuation, according to CNBC data cited in the episode. The board also retains the right to add another 2% whenever company valuation increases by 50% over an undisclosed baseline. Elon Musk immediately responded on X, arguing that instead of placing shares in a fund, distributing them directly to ordinary people would be better, and announced that starting next year, SpaceX will give one share to each American newborn annually as the beginning of similar programs.
— Lon HarrisDirect stock distribution beats charity foundations for public benefit
In an old clip from April, Jason argues that American hostility toward AI stems largely from the public not sharing in its gains—people holding VC fund stakes or 401(k) positions in Nvidia or Google feel positive about the economy, while those without equity feel left behind. He contends that instead of funneling money into traditional charitable foundations that consume most management fees, distributing shares in companies like SpaceX directly to ordinary people would truly shift overall public sentiment about AI.
— Jason CalacanisPublic benefit funds can become opaque slush funds
Jason expresses skepticism about Anthropic's public benefit fund itself: once money enters the account, management decides how to spend it rather than distributing it directly to the public, creating ample room for operations—it could become a vehicle for taking care of insiders or generate new group antagonisms during distribution. His conclusion is that this money carries the imprint of management and will likely become an opaque fund, which is why he favors the direct stock distribution approach.
— Jason CalacanisRobot muscles use hydraulic contraction, not motor-pulled cables
Clone Robotics' technology path differs entirely from Tesla Optimus, Figure, and Boston Dynamics. The latter all use motors to pull cables simulating muscle effects, essentially pulleys and cables at their core. Clone invented hydraulic tendon units by pressurizing an elastic tube wrapped in braided sleeves; the braiding angle constrains outward expansion, forcing contraction along the length axis, directly producing force similar to real skeletal muscle contraction rather than simulating it with motors.
— Dinoosh RadhakrishnanHydraulic muscles claim one-tenth the cost of motor drives
Dinoosh attributes the biggest advantage of hydraulic muscle technology to supply chain economics: motor-driven humanoid robots depend on many custom gears that are expensive in themselves, and every company competes for the same suppliers—an industry bottleneck. Clone's design requires no gears whatsoever; according to him, even without economies of scale, it already costs one-tenth as much as motor-driven humanoid robots. He believes this cost advantage, not fidelity itself, is the core advantage venture investors will ultimately care about.
— Dinoosh RadhakrishnanRobotics firms bet on free, powerful open-source AI models
Clone has spent nearly all its capital over the past five years on hardware and artificial muscles—a problem no one else is tackling—while treating intelligence as an explicit wager: Dinoosh believes frontier models will be replicated or open-sourced at cheaper cost within six-to-twelve-month half-life cycles, citing Nvidia's GEAR project. Once real robot data exists, with sufficient capital they can take the latest open-source models and apply reinforcement learning to remain competitive with frontier capabilities without training large models from scratch.
— Dinoosh RadhakrishnanEnterprise robots deploy first; home assistants wait two more years
Dinoosh provided a concrete timeline: by early 2027, first deployment of enterprise-grade robot torsos capable of dual-handed tool use—drills, hammers—for assembly lines and fixed-location scenarios; he believes this sector can generate tens of billions to hundreds of billions in revenue. Full home-use robot butlers will require another two to three years. On pricing, he endorses $25,000 as a reasonable target price, while noting they will eventually produce million-dollar luxury versions.
— Dinoosh RadhakrishnanElevenLabs cloned a podcaster's voice for its Archer model
Chris Williamson, host of Modern Wisdom podcast, accused ElevenLabs' Archer voice of being a clone of his own voice and released comparison clips. ElevenLabs told the public it hired a separate voice actor but provided no public evidence, meanwhile quietly removing Archer from the platform. Jason's judgment is direct: this is stealing voices caught red-handed; the company should compensate and formally purchase his voice license—something ElevenLabs already has as a legal option and has previously done with celebrities like Michael Caine.
— Jason CalacanisIn their own words · checked verbatim
Just give people directly some goddamn shares of SpaceX and the entire problem is solved.
Jason Calacanis26:29
That to me, he's got a lot of fingerprints on it that it's going to be a slush fund.
Jason Calacanis30:36
But at Clone, we invented a hydraulic musculotendon unit. It's literally just a muscle tendon. So you pressurize our muscle fiber with fluid, whether that's gas or liquid, in our case, liquid.
Dinoosh Radhakrishnan43:58
is that it can be done at a tenth of the cost to produce of these motor actuated humanoids.
Dinoosh Radhakrishnan56:20
on the intelligence end of things that the models would end up having like six to 12 month, like half-life in effect.
Dinoosh Radhakrishnan58:20
It would be as early as, you know, Q1 2027. We'll deploy our first silos with us.
Dinoosh Radhakrishnan1:03:30
if you are a fly, simulation theory is now real.
Lon Harris1:11:53
It is 100% based on his voice.
Jason Calacanis1:23:06
Figures
| Anthropic public benefit fund equity value | ~$100 billion at $2 trillion valuation | 20:26 |
| Additional equity per 50% valuation increase | 2% | 20:26 |
| Annual U.S. newborns | ~3.6 million | 21:26 |
| Clone robot cost vs. motor-driven path | ~one-tenth | 56:20 |
| Clone Robotics total funding (5 years) | $24 million | 1:00:23 |
| Clone enterprise robot deployment timeline | Q1 2027 | 1:03:30 |
| Odyssey fruit fly brain video views | >5 million | 1:11:53 |
Glossary
- Public Benefit Corporation (PBC)
- A company whose charter legally enshrines pursuit of public benefit as a corporate mandate.
- McKibben muscle
- A classic pneumatic or hydraulic artificial muscle using a braided sleeve wrapped around an elastic tube, contracting axially when pressurized.
- Antagonistic actuation
- Using paired flexor and extensor actuators that pull against each other to control both directions of joint motion.
- Preference stack
- The phenomenon where celebrity support shifts wholesale with political winds, one allegiance replacing another.
How to listen
Founders and investors tracking AI company equity and governance design, humanoid robot technology choices, or voice-clone copyright risks.
The opening ~20 minutes of movie reviews and casual chat have low information density and can be skipped.