Tim Smith retweeted
@gearsdotco will make billions of gears and gearboxes in America!
Thank you @jakobdiepen and @Chasecaffrey, the @DiscipulusVent was an amazing experience 🫡
Also thanks to @shaunmmaguire @isaiah_p_taylor @ADoricko @jimbelosic for spending time with us, sharing your knowledge and inspiring to move fast
Gears (gears.co) — Gears for robots in America
Gears account for 30% of a robot's bill of materials, and foreign parts are facing bans. @alexpodkin is solving the gear shortage at home. Shipping gears in 3 weeks.
📩 [email protected]
Tim Smith retweeted
I started using the gitea ticket flow over the weekend on a personal project, and it's great. Much closer to how I do things at my day job.
I think the biggest benefit is grounding/observability of the agent output. It reminded me of @dirtman 's post.
Tim Smith retweeted
I do not support pacing frontier models. If I am going to die at the hands of killer AI, I want it to be American, not Chinese. Buy American, Die American.
Tim Smith retweeted
You may laugh, but my car has never been so precisely docked in my garage.
This object Orion is looking at is an International Docking System Standard (IDSS) Centerline Docking Target (CDT), Rev E. It's not just cool looking, there is a lot of information embedded in the geometry that a pilot can use. Print your own here!
printables.com/model/1430555…
Tim Smith retweeted
The “company” on the left is a nation state
If I were the company on the right I would try to kill every company that is like the one on the left.
This way, the company on the right can make trillions of dollars and then infect American politics with hundreds of billions of dollars to implement their vision of being the sole judge, jury and executioner of future progress.
Tim Smith retweeted
For drone startups, I quote them "crazy eddie" low prices for swiss lathe machined parts because I love what they are doing and want them to win big I don't even care about making a profit honestly we need to get as many startups off the ground in this space as possible.
Tim Smith retweeted
Humanoid robots will probably have a huge place in our future (especially on distant and dangerous missions, like Mars), but they will definitely not take over the factory floor.
We developed a bench top robot arm that can move its end effector to any point within ~5um inside of an 800mm work area and the BOM for a single, prototype unit is under $1,000. That falls below $400/ea with moderate scale (low hundreds of units/year). There will never be a humanoid robot that can outcompete this architecture on price, speed and precision.
The cost of raw materials by weight is already off by an order of magnitude in China. In the US, it’s closer to 100x.
There are many other issues plaguing the humanoid form factor (like bipedal locomotion), but price and complexity alone are enough to make them an obvious non-option for any case where repetitive, similar motions are needed in a relatively small area (i.e. manufacturing).
The price for precision actuation is going to fall through the floor in the coming years and humanoids have zero slack to keep up.
JUST IN: A new McKinsey & Company report breaks down exactly what it costs to build a humanoid robot, and why the supply chain is one of the biggest bottlenecks nobody is talking about.
The typical humanoid bill of materials today: $30,000 to $150,000 per unit.
The long-term target to unlock mass-market demand: under $20,000.
That's a LOT of cost compression still required.
Here's where the money goes:
→ Actuators — 40-60% of total cost, and the PRIMARY performance differentiator
→ Sensing & perception — 10-20%
→ Compute & control — 10-15%
→ Structure — 5-10%
→ Battery — 5-10%
The uncomfortable truth: the most expensive component, actuators, also has the LEAST developed supplier ecosystem.
And here's the scaling dilemma nobody has solved yet. Suppliers won't invest in dedicated production lines because volumes are too low.
But volumes stay low because costs are too high. A classic chicken-and-egg problem.
The one structural advantage? China. Its deep EV supply chain overlaps directly with humanoid components, motors, power electronics, permanent magnets, precision bearings.
That's why Chinese manufacturers have a significant head start on cost curves.
Western humanoid companies are racing to either vertically integrate or lock in co-development partners. Neither path is cheap. Neither path is fast.
Everyone is excited about the humanoid robot race. Not enough people are talking about the supply chain war underneath it.
That's where this gets decided.
P.S. It's good to see Schaeffler providing 'picks and shovels' in this humanoid race. 🇪🇺
McKinsey Report here: mckinsey.com/industries/indu…
~~
♻️ Join the weekly robotics newsletter, and never miss any news → ziegler.substack.com
Tim Smith retweeted
Personal rule of thumb: don't use an LLM for something that a deterministic program can do.
I get it, LLMs are exciting, but they don't mean that software ceases to exist. They are fantastic at dealing with human language and ambiguity, but are terrible (by design and for good reason) at repeatability.
To borrow terminology from the book Thinking Fast and Slow, LLMs are "system 2"...slower, more "expensive" (for LLMs, both in time and dollars), but flexible and creative. Traditional programs are "system 1" ..fast and cheap, but inflexible and dumb.
Instead of trying to put an LLM in the "hot loop" of your program, it's usually worth asking an agent to write a deterministic program to do the thing you need done. Since code is "cheap", this deterministic tool can do exactly what you want it to, and doesn't consume tokens on every execution.
(This applies to agents too..I find myself regularly yelling at Claude to stop repeatedly generating the same 30 lines of python to inspect a file, and instead telling it to generate a 3-line shell script wrapper around jq that it can check in and call repeatedly)
Tim Smith retweeted
Cubemars budget: $0. Kinematics analysis: 100%. Designed these AK80-64 replicas to save $10K+ while prototyping. The heavy-duty bearings roll with zero output shaft deflection. Prototyping smarter, not harder. 🛠️🦾
#Hardware #Robotics #MechanicalEngineering
Tim Smith retweeted
Previous experience of my newest 12 manufacturing hires:
- donut shop worker
- accounting assistant
- elderly caregiver
- janitor
- bartender
- AI data analyst
- call center rep
- nurse at the VA
- associate at Dollar Tree
- hospitality associate (?)
- school custodian
- graphic designer
I'm asked all the time about "how will manufacturers find skilled labor as we scale and reshore" and the answer is we have to create it ourselves. Relying on someone else (a school, the government) to create skilled labor for you means you're gonna be waiting a while. Just do it yourself.
On-the-job training is the only way I've ever known how to hire. I think I was really lucky to come from a small town where OJT was common, and often the only way to learn how to do something.
We have a 19 year old running a $1M Matsuura MAM, and a 70 year old running another machine next to him. Both produce great parts. The labor supply is endless if you are willing to put in the effort.
Tim Smith retweeted
1. You have a problem that needs solving.
2. You solve it and it needs building.
3. You build it and it needs testing.
4. You test it and it needs refining.
5. You refine it and test it again.
6. You test it again and encounter new problems.
...
101. You shine it up and ship it.
Tim Smith retweeted
MEN WANTED for hazardous journey, small wages, bitter cold, long months of complete darkness, constant danger, safe return doubtful, honor and recognition in case of success
THE ROBOT AI HYPERWAR IS COMING
WE ARE BUILDING THE INDUSTRIAL MACHINES TO WIN THE ROBOT WARS
DO YOU LIKE POWER ELECTRONICS, CNC MACHINES, ROBOTS STONG ENOUGH TO CRUSH A MAN AND LASERS THAT CAN MELT STEEL?
CAN YOU COMMAND THE FORCES IF NATURE TO BEND TO YOUR WILL?
APPLY BELOW! 👇
Tim Smith retweeted
These are great tips and accumulating any of these skills is definitely better than not having them.
For new players out there, please don't believe that you need to attain every bit (or any bit) of knowledge on this list in order to make something useful.
If you're (like me) not the kind of learner that can grind out years and years of prerequisites, hoping to eventually know enough, start with the videos and tutorials.
Make some hot garbage on your own time.
Get some burns and cuts and scars.
Make something that has some tiny bit of utility and then make it better by improving one area at a time.
The best time to get started was yesterday. The second best time is today.
Tim Smith retweeted
Please sell me your manufacturing startup’s product here
Let’s make a thread of manufacturing products and services
Tim Smith retweeted
> Theyre solved.
The supply chain isn't solved.
Making things cost-effectively, at scale, in the USA is a hard problem, and drones encapsulate a collection of critical supply chain sub-problems:
- brushless motors
- energy dense batteries
- electronics/sensors
- light weight chassis
It's great that @PalmerLuckey and @elonmusk have provided proofs of concept that these things can be built in the USA (again), but output needs to be 10,000x what it is currently, and there are reasons why it isn't, and those reasons represent Hard Problems that need Solving.