Press tools for car body panels are extraordinarily expensive, which is why low-volume manufacturers generally avoid using pressings wherever possible. It's just inherently very expensive to carve two huge blocks of steel into a smooth curved shape, so you need to sell an awful lot of units to amortise that cost. Tesla's deal with Lotus only worked because they used a fibreglass body - expensive per-unit, but very low tooling costs.

Desktop Metal are developing a sheet forming solution that requires no bespoke tooling, but it's a slow process with fairly poor surface finish.

https://www.youtube.com/watch?v=6oqeVLILGHY

▲

cduzz 4 days ago | parent [-]

Right -- this is why I use the analogy of TSMC -- chip fabs are also extremely expensive, for similar reasons.

What are the relative costs of the making die set, the press, and setting up and doing a run of stampings, and the facility and employees to actually house the whole kit?

As of right now, if you need to make a car and you don't have a NUMI or similar retired automotive plant sitting around, it's going to be expensive.

What about the hydroforming process?

I guess smaller car makers from the 60s that did make low volume sheet metal cars didn't need to pass crash tests.

Probably The Telo people should just team up with ineos....

▲

numpad0 2 days ago | parent | next [-]

The process of a 3D model -> dies > panels that match the model is iterative. Tesla apparently cheaped out on that optimization paths and they had lots of issues towards rear left corner of cars, somehow on both Model S and X.

ODMs like Magna-Steyr and Valmet also exist. They take your plan, build some, and send to you on a ship from Central Europe.

▲

convolvatron 4 days ago | parent | prev [-]

except the fab pays the tooling cost once per process and the tool and die company charges once per design.

I've actually worked a little on hydroforming, but unless you're thinking of a different kind, it was labor intensive, and prone to crinkling in bad spots. we basically concluded that we used less time and got a better result with an English. which would probably run at least $10k for a car body, if you could find someone willing to work for that low an hourly rate.

	▲	cduzz 4 days ago \| parent [-]
		Again, I'm confident that this is not a viable business, and certainly in no position to make it a viable business even if it was... But places that make windshields keep forms around and make runs of windshields even for low volume cars; obviously there's less recurring need for fenders or floor pans... But there may be some way to financially engineer a "do a run we pay X per unit; hopefully that results in doing another run and paying X, and if we can get to 100X the per unit costs go down to some other target. But I was thinking, and perhaps they should just make these cars out of ground up trabants or saturns. Tesla has demonstrated that some customers don't actually care about perfect finishes or gaps or whatnot; I certainly would be fine with a car made out of trabant, especially if it meant I didn't have to worry about dents...