In fairness, the market size isn't in question. It's whether you can reasonably capture venture-scale share in an already-mature market.
They need to see some kind of unlock. In the case of SCS and Nox, it's a new business model, enabled by customized SW, and aided by a generational shift in buying behavior, and an ongoing wave of legacy retirements that are creating a capacity vacuum.
Those factors convince investors that perhaps now, finally, manufacturing demand can be aggregated and the supply chain reshaped pretty fundamentally. That thesis is backed by their rapid growth, and the rapid growth of competitors in the same and adjacent markets.
Electronics grew up with tech (for obvious reasons), so it's less obvious that something dramatic can happen to reshape the supply chain.
If you are just another competitor in an already-consolidated, tech-driven marketplace, what's the venture bet?
Oh look, a full-scale Saturn V rocket made out of OSH Cut parts. Just $1.77M.
Can your sheet metal vendor quote something like this? ;-)
The temptation to be everything for everyone kills the best businesses. Don't compete in markets where you aren't special. Don't offer services that don't fit your model. Keep your cost structure low and your operations simple.
Caleb (OSH Cut) retweeted
Big trophies for an even bigger event. 🏆🔥
We teamed up with Matt’s Offroad Recovery to help bring the 2026 Smash & Dash trophies to life, and our tube laser made quick work of them.
I don't know what's going to happen to CAD software. But Claude Code downloaded Python and OpenCascade and just designed what I wanted on its own.
Probably nothing.
Phrases to watch:
- "Closed-loop, manufacturing-constrained generative design"
- "Manufacturing execution layer for AI-generated hardware"
- "Infrastructure connecting AI design to physical manufacturing"
- "Manufacturing infrastructure for autonomous hardware design"
Oh, hey! xitter.cf/OSHBuilt/status/210767…
This got me thinking...
so the productized lanes @OSHBuilt wrote about last week have a 28-year history. i did a bit of digging this week. here's what each one said NO to:
> protomold, 1999: anything that wasn’t a two-part aluminum mold
> quickparts, 2000: no quote without a file, and they patented that
> osh park, 2009: one purple stackup, $5 a square inch
> rapid sheet metal, 2013: no prints, 3d files only (sold for $120m)
> sendcutsend, 2018: no prints, 44 by 30 inches max
> osh cut, 2018: same price for everyone, no negotiation
> fabworks, 2023: metals only
> digital metal, 2025: no tooling, castings from a printed tree
> finalrev, 2025: step files only, same day
> twentyfour26, 2026: one part size, 6061
> phoenix heat treating, 2026: in testing
there are more for sure, but many didn’t survive (plethora, tempo, fast radius, shapeways), mostly because they said YES to everything.
but yeah, this is a long one. give it a read over the weekend and let me know who i forgot or got wrong.
THIS JUST IN: "Supercar not practical," stunning insight from unexpected corner of the internet.
Caleb (OSH Cut) retweeted
Have you ever wondered what OSH Cut's operators are thinking while manufacturing your parts?
This is a reverse-dox. All we know is this guy isn't a controls engineer.
This is a deeply impressive demo, but because of the design and integration problem, and the chutzpah. But the controls problem isn't that complex and doesn't require that much compute.
Hate to dox myself a bit but have worked on just about every fighter jet the US has. I have seen some impressive technology over my career but this is a whole other level. I have so many questions and need to know how this works. How much edge compute is needed to make this happen? My first thought is dual processor + 1TB ram. That way it can handle the fuckquintillion process per second. There was zero correction it nailed it 100%.