To give some sense of the difficulty in machining to that precision: the difference in a 4" block of steel at 68 degrees vs 74 degrees F is about 5 microns. Everything gets MUCH MUCH MUCH harder as you get bigger if you want to maintain precision. Truly an amazing feat of engineering.

How is the thermal expansion coefficient of steel relevant? Wouldn't you use quartz for the mirror, which has an expansion coefficient more than two orders of magnitude smaller?[1] The James Webb telescope has a 6.6m wide primary mirror, but it's actually composed of 18 1.3m hexagonal segments.[2] (Unfortunately, I can't find any sources for the material used.)

The Hubble telescope was launched in 1990. It was, presumably, at the bleeding edge of modern engineering. 30 years later it shouldn't be unthinkable that commercial engineering could achieve something comparable at a fraction of the cost, not to mention all the improved methods for compensating for defects and utilizing smaller, more easily manufactured components. Similar advancements are what have made cheap launches possible.

Yes, the engineering required for these things is still amazing. But that doesn't imply it's still as expensive. Plus, there's more private wealth. Maybe crowd sourcing isn't practical, but I'd think universities could easily achieve this if the motivation was there. Nobody thought launch costs could be reduced as much as SpaceX has achieved. All it took was a highly motivated person. Musk didn't invent any new technology; like many industrialists he simply recognized the technology was already there or at least on the cusp, and assembled the assets to make it happen.

[1] http://hyperphysics.phy-astr.gsu.edu/hbase/Tables/thexp.html

[2] https://jwst.nasa.gov/content/forScientists/faqScientists.ht...

Optics technology is not advancing that fast. It had a rapid growth phase in the 17th and 18th centuries, and then a lot of the techniques remained almost unchanged until the mid-20th century, when the laser improved a lot of things. But consider that the big spectacular innovation of the last few years is Rayform, which patented a gradient-descent algorithm for making a makkyo mirror, as people have been doing for thousands of years.

There are some advances in, for example, using glass-ceramics instead of glasses, which give better strength, better rigidity, and enormously better fatigue tolerance. I don't think anybody makes multi-meter mirrors out of fused quartz, even though it would be a near-ideal material from a TCE perspective.

That's a great way of comparing it. I had no idea the hubble error was so small.

It's very difficult, but it's quite common for amateur astronomers to grind mirrors to a precision of 0.1 microns or better, using century-old techniques. (Typically the mirrors are smaller, but that's a matter of the available budget for materials and man-hours, not the metrological precision.) I think it's reasonable to assert that if a company doesn't have such amateurs on staff, it will not be able to fabricate optics to such demanding precisions. There are not currently any assembly lines for such things.

The century-old techniques are a little better now that you can use a laser instead of a candle, and there are improved techniques you can use, but amateur telescope fabrication techniques are totally capable of hitting λ/8 precision.

The difference is size, the simple grinding process does not keep accuracy for big mirrors and lenses.

The accuracy isn't in the process of grinding; it's in the process of measurement. All the grinding process needs to do is to remove only a small amount of material, which is itself a nontrivial problem; a single large grain of grit can ruin weeks or months of work, so amateur astronomers may get very angry at you for things that seem unreasonably minor, like opening a door without washing your hands. But that's not a problem that scales superlinearly with mirror size.

Really large mirrors are segmented anyway.

Humble brag on my alma matter which has one of the worlds largest mirror labs underneath the football stadium. Cool 'Tom Scott' video of the process involved in the glass processing: https://www.youtube.com/watch?v=BP9HNVuGb-g

I'm hopeful that adaptive optics and computational imaging techniques can help here