You're confusing the complexity of the product itself with the complexity of being able to produce said product at scale to acceptable levels of quality. These are not the same thing.
The amount of naivity around this topic is actually scary. Seems like a lot of people really do assume every physical/ made by hand product is easy, and requires no talent or experience. Meanwhile, 90% of the posts on here about tech products end up somewhere along the lines discussing dunning Kruger, and false superiority. The dichotomy is wild especially knowing that when people burn out in tech they often end up working in a skilled trade.
Physical products are hard. Automating them means you've screwed up half a million times and have a lot of cash to even do it.
This! Although I think the most likely failure considering the amount of money spent on the plant is that they probably dove head-first into automation and hit the integration challenges this can have.
(Perhaps a business case of “on-shoring makes sense if everything is automated!” that failed to account for the challenges of opening a fully-automated facility)
Ever see that meme about newcomers to ML make a giant leap to bert/chatgpt? Id wager 100 bucks you're right. Also bet a lot of dudes nephews made a lot of money while being unqualified for their positions.
You have to make those procedures first, and just like software, manufacturing technologies change fast. Each manufacturing technology has its own bugs and a lot of older technologies you can't even hire for anymore.
Now get this, someone could say "we're going to do it with AI from day one!" Blow 40million, to contractors because ew who wants to own their production, and fail. Black box technologies and contractors have different risks entirely.
There's a lot of reasons why attempting to revive thirty year old practices to still be profitable is more than a challenge. Hence why they failed.
I don't buy it. I think what happened is they lead with the budget and timeframe. What good does that do? You get 90 million and 5 years, are you going to actually implement the thing? No you are going to get the perfect hybrid of political lawyerly parasite to suck the 90 million dry and the short timeframe makes it more attractive to the parasite not less.
The explanations make no sense in the article. Automating heat treatment is computationally expensive? What the hell are they talking about? The guy who is doing the heat treatment is not doing computation!
Oh yes who needs to sense and regulate the heat in a place forging, annealing, and quenching various metals while keeping both people and property safe! Also who would want to save money because the energy to melt tons of steel is very cheap.dont even get me started on the HVAC system! It's not like those are prime targets in lesser demanding industries for automation or anything.
But they're very closely related. A complex item is going to have a more complex manufacturing process than a simple item. Have you ever seen how big an aircraft factory is?
>It took Airbus 21 years before they made a good airplane. It took Tesla 15 years from start to significant production volume.
Extrapolating from complexity it would only take a few months at worse to make a factory producting drop forged box end wrenches.