Possibly, but he also describes pretty cheap and easy calibration methods. It's possible that most of the cost of the commercial options is the fact that these are extremely low-volume devices and the overhead of a business with low-volume sales is quite high. You might be paying less for calibration and more for all the things a business needs to do/have that a lone DIYer doesn't.

Not to mention the fact that a lot of researchers are buying thing with grant money, and so can be, in some cases, somewhat price insensitive.

I don't doubt that the commercial one is pre-calibrated and certified. But I would be quite surprised if that certification/calibration _actually_ cost ~$9000.

DIY is not just for hobbyists.

Companies face the build-or-buy question every day. This article demonstrates that you can build and validate your build with the knowledge and skills expected of most principal investigators. If I were a lab director, I would want my people to be able to consider building when it makes sense, particularly since we can then integrate systems and tailor validation to our requirements.

That's the stuff of proprietary IP and career advancement.

The point wasn't "you can get rich selling this", but "you can save $$$ building this yourself as a hobbyist".

I'm not sure about certification but there is a lot of overhead in designing, producing, selling and servicing products. This is basically a prototype. It will probably work for a while but will be difficult to maintain if something goes wrong and the software won't be very polished. There are a lot of non-BOM costs that go into selling a commercial spectrometer. This is great if you want to DIY one but even a single sensor based visible light spectrometer with cheap parts can't be bought for close to the sum of its parts.

If you're using it in a lab, you can calibrate and verify this sort of instrument yourself most of the time. Wavelength is trivial to calibrate. Absolute amplitude not so much, but you usually don't care that much about absolute amplitude. Response flatness across the spectrum may or may not be a concern.

So what? If you need it, now you can build it. As a hobbyist you don't need certified gear, you need gear.

I wonder what 3rd party certification would cost. Industries that use sensors like this frequently require annual recertification, so there are typically multiple testing houses and competitive prices.

My first thought as well. People tend to misunderstand why things cost what they do, for stuff like this it isn't the raw components.