Building high-mix high-volume future with robots and AI @machinalabs_
Joined October 2021
- Tweets3.5K
- Following453
- Followers12K
- Likes6.1K
Pinned Tweet
(1/9) @MachinaLabs_, we're working on full automation in manufacturing through our 'Robocraftsman' technology—robotic cells designed to turn design intent directly into physical products. This concept is akin to the shift in the digital world, where cloud computing allowed software developers to focus on creation without the constraints of physical infrastructure. That shift enabled the rapid expansion we see today, notably the rise of artificial intelligence and AGI on the horizon.
To enable the same rapid speed of innovation in physical world, our team is building a similar abstraction layer in manufacturing, making it as effortless to produce a new product as it is to launch a web service.
Initially, I was reserved about sharing the details from our internal all-hands, as they capture the core of Machina’s ambitious goals. However, I realized the value of opening up our vision to the wider community. We’re on a mission to transform manufacturing, and it’s a journey that extends beyond just one company. We invite entrepreneurs and innovators to join us in reshaping this fundamental human expression, manufacturing.
I think we gonna open up a new role at Machina called “Vibe Coding Sherpa.” To help all the team members who are vibe coding to become better at it. Current experienced software engineers will be future vibe coding sherpas. Who said AI kills jobs?
Hey @grok; Which US manufacturing startup companies deserve to be on this but are left out and why? :)
The 2026 American Dynamism 50 is here: our annual list of 50 companies, from within the a16z portfolio and beyond, building for America's future.
This year's theme is Production is Deterrence. "America's ability to prevent conflict depends on what it can build, how fast it can deliver, and how long it can keep going."
A new generation of defense companies is moving from prototypes to full-scale production, and the industrial base has to keep pace. This year's list is organized around three pillars:
1. Systems: the drones, ships, missiles, aircraft, and spacecraft that deter conflict.
2. Factories: the plants, tools, and software that build them.
3. Inputs: the materials and power they're made of and run on.
Get all three right, and the impact reaches well beyond defense, to skilled jobs, revitalized communities, and the supply chains behind energy, housing, and everyday goods.
Meet all 50: a16z.news/p/ad-50-2026-produ…
We love building for defense. We’re just as excited about being a bad influence on car designers.
Roboforming opens the door to entirely new commercial product categories. We’ve been working with some of the best in automotive to show what that makes possible.
Unveiling at the LA Auto Show.
We partnered with @DesignsCallum to introduce Lassen at this year's @LAAutoShow.
World-class automotive craftsmanship meets a factory without constraints.
See it November 20–29.
#LAAutoShow #MachinaAutomotive #IanCallum
“Be afraid. Build factories.”
That’s too much of today’s reindustrialization pitch.
Yes, defense has a production problem. But making production accessible has a much bigger promise: a world where anyone with an idea can actually build it.
Want to excite the next generation? Show them what they could create, not just what they should fear.
Build for that. Deterrence comes included.
I don’t even know why you would wanna be a founder if you don’t wanna be into technical details. That’s the only fun part 🤣
Some of the most novel manufacturing research right now looks like a blacksmith shop connected to a GPU.
A few things happening in labs that folks here are not aware:
AI-generated forging sequences are running physical machines. Simulating how to deform a billet, generate a sequence of operations, and test it on a real forge. Initial demonstrations are already happening.
English wheels are getting robotic control and machine vision. Combining shape measurements with models that predict how the next pass will deform the sheet.
Metal printing is being combined with forging and hot peening. These operations change the material’s internal structure as well as its shape. Experiments on printed aluminum have already shown improved ductility after forging.
What would prove manufacturing technology has fundamentally advanced?
Throughput alone is too low a bar. More output can simply mean more capital.
My candidate: how many genuinely new designs did we (as human race) take to 1,000+ good units?
The deeper, Thielian question: what can we manufacture economically today that we couldn’t before?
Libertarianism had a staffing problem. AI + robots might fix it.
Tiny groups. Huge power. If everyone gets distributed access instead of premium subscription to a central tool.
Also, when anything can be faked, your inner circle becomes everything.
Intelligence gets cheap. Trust and loyalty becomes super valuable. That would probably be one of the few good truly human attributes left.
What would prove manufacturing technology has fundamentally advanced?
Throughput alone is too low a bar. More output can simply mean more capital.
My candidate: how many genuinely new designs did we (as human race) take to 1,000+ good units?
The deeper, Thielian question: what can we manufacture economically today that we couldn’t before?
The beauty of technology is that it compounds. The first few building blocks are painfully hard to create. Then you have enough, and suddenly you’re building things nobody thought possible.
That’s why this is a smart move by Anduril. On the surface, they’re selling computers. The bigger payoff is a low-cost electronics foundation that makes everything they build on top of it cheaper.
@PalmerLuckey knows what he’s doing.
America’s manufacturing problem isn't capacity. Turning an idea into physical is still too inaccessible for most.
The tech that wins will turn “someone should build that” into “I built that.”
That’s why we have Roboforming Fridays at Machina. Even our nontechnical staff design and build metal parts with Robocraftsman.
We’ll reindustrialize when people feel as capable of building physical things as AI is making them feel about building software.
Today, our team. Eventually, everyone.
I’m less interested in whether the new Roadster is practical than in the racing league you could build around it.
Who’s starting this? I want to invest.
This might be engineering’s most misunderstood photo this year.
I believe @davidsenra and @tobi discussed it as an example of engineering’s ultimate goal: removing complexity.
But looking simpler and being simpler are two very different things. Sometimes complexity gets eliminated. In this case it went out of sight.
That’s the magic of 3D printing. You can pack an astonishing amount of complexity into a part that looks beautifully simple. The camera doesn’t get the full story.
Compare the first Raptor to the latest and it looks like parts disappeared. They didn’t. The complexity moved inside, and advanced manufacturing made that possible.
@EdwardMehr and @ErikPalitsch on what people miss about the value of 3D printing. cc: @MachinaLabs_
20+ products in 8 yrs. Love the ambition. SpaceX spent 26yrs with only: Falcon, Dragon, Starlink, and Starship.
Production would be a big challenge. Making mid volumes across this many complex products is its own engineering problem.
I’m sure Anduril has something up its sleeves. Hopefully a few manufacturing breakthroughs.
A first pass at ranking manufacturing startups using public information and estimated inputs; not audited financials.
I’ve replaced the names with their operating personalities. You can guess who’s who :)
America’s manufacturing gap is the middle: hundreds to thousands of units per design, per year.
Too much for prototype economics. Too little to spread fixed facilitization costs across millions of parts.
There is no metric to show which company could win. VCs have SaaS metrics. ARR, TCV, retention don't clearly show anything.
My proposed metric:
Resilient Capacity Velocity =
(New qualified capacity / time) × flexibility × demand buffer
Qualified capacity: sustained output at the required quality and batch size.
Flexibility: How fast and cost effectively the company can switch to a new design.
Demand buffer: how far revenue can fall before operating profit reaches zero.
Measure how fast companies build capacity that can survive a change of plans. Highest scores will win.
America’s manufacturing gap is the middle: hundreds to thousands of units per design, per year.
Too much for prototype economics. Too little to spread fixed facilitization costs across millions of parts.
There is no metric to show which company could win. VCs have SaaS metrics. ARR, TCV, retention don't clearly show anything.
My proposed metric:
Resilient Capacity Velocity =
(New qualified capacity / time) × flexibility × demand buffer
Qualified capacity: sustained output at the required quality and batch size.
Flexibility: How fast and cost effectively the company can switch to a new design.
Demand buffer: how far revenue can fall before operating profit reaches zero.
Measure how fast companies build capacity that can survive a change of plans. Highest scores will win.
First-principles manufacturing starts with a simple question: What is the shortest path from raw material to a useful product, with the least waste and complexity?
Dedicated tooling IS waste. Machina and Freeform are attacking that constraint directly through Roboforming and 3D printing. Eliminate part-specific tooling. Only then you relentlessly improve speed and throughput.
“Eliminate before you optimize” was the big lesson from SpaceX.
Old/current manufacturing: make every box in the flowchart faster.
Next-gen manufacturing: delete boxes.
Flexible, hybrid machines mean fewer steps, less complexity, more people able to build.
As @ErikPalitsch said in our conversation. It is incorrect to say "reindustrialize" America. Neoindustrialize it!
What does a factory look like when the boundaries between operations disappear?
Sensing and control span every step. Parts move as one continuous process. The factory is the machine.
@ErikPalitsch and @EdwardMehr dig into it at @MachinaLabs_.
Full discussion👇🏽
You can’t "reindustrialize" America with SpaceX cosplay and an AI bumper sticker.
Erik helped build early SpaceX and now uses AI without much hype to develop new manufacturing processes. If we want the next generation of American industrial giants, more founders should study his playbook.
America doesn't need to rebuild the factories of the 1950s. It needs to invent the factories of the future.
@EdwardMehr and @freeform_future CEO @ErikPalitsch discuss reindustrialization, lessons from SpaceX, outmaneuvering China, and the future of physical AI.
Full discussion 👇