And service robotics in last mile and distribution is going to be much bigger than manufacturing. Consumerization of robotics.
Conversation
Replying to
Yea totally, but I'm saying that there aren't many big robotics companies here to lobby for these gov't measures. I suspect it'll be more organic, but I hope I'm wrong.
1
1
Replying to
Yeah I agree. I think there’s a cost inflection point coming, like 2000 for software startups. Cost of robotics startups should plummet, as some of the AI gets dirt cheap and hardware gets commodity/open-source. Trigger is renewables plus falling lithium-ion costs.
1
2
Replying to
I think a lot of the push will come from green. Same tasks at lower, greener energy cost. Delivery, distribution, simple service. Human operators run on carbon-based energy. AI is more near-zero-marginal-cost enabler after initial training costs. Training = capex. Energy = opex.
1
1
Replying to
Electric power is already cheaper than caloric power though, so while power density has real practical implications, I don't think electric costs are much of a real barrier. Brains are harder than muscles (not to say robot hardware is solved, far from it)
2
1
Replying to
Point of use price is not the same as production price. Non-grid-rechargeable batteries are like orders of magnitude higher per watt and still sell briskly. They’re like photographic film in ~2000. I think of “electrify everything at pennies/kWh” as an unfinished 120y old project
2
1
1
Replying to
Numbers don't seem to work though. Maintenance / ops is generally way more than power cost in true robotics (ie not high power machine tools) if we're talking unit cost.
This is true of CapEx and OpEx.
1
1
Replying to
Remember it's maintenance and ops, and the latter could be a big problem unless telepresence get really advanced.
It's unclear how good maintenance can get. Cars, computers, etc. are at very high unit volume while still being expensive to maintain.
1
1
Replying to
I am too. Cars and 6 sigma have been around for a long time. Sigmas likely run into transistor-size style physical constraints, requiring new modes of automated maintenance, which is possible, but we'll see.
1

