It depends on what you mean by mainstream and by large company but Clojure, Elixir, and Elm are all languages that are frequently on HN and they were either personal projects or backed by small companies (<40 employees). I think Julia too.

Clojure was a lonewolf in the wild, no backing from any big company for a long time (that's if you don't count Cognitect). The largest fintech in Latin America, NuBank is now behind it.

elixir and zig seem to be good candidates for going mainstream

Crystal has been getting at least some funding from various companies. Maybe not what you would consider a large one, though.

Companies love opportunities for sponsorship to get promotion (and status) back. That means unless ideologically opposed almost all small open-source projects will get a larger company to backed them up. I think there are more direction and governance issues than financial.

Julia

It got MIT backing, no?

MIT doesn’t fund anything — that’s not how academic money works. The university doesn’t give out money, researchers apply for grants from the government, companies, or NGOs. The university then TAKES a huge cut of that money — it’s called “overhead” and it’s often more than 50%.

Initial Julia work at MIT was funded by some absolutely tiny grants for distributed linear algebra research. The total amount of funding that Julia has gotten via MIT is definitely orders of magnitude less than the salaries paid to Swift, Go and Rust developers by Apple, Google and Mozilla.

My experience in European universities is a bit different, yes scholarships are relevant, but salaries also happen regardless of them.

I can tell you first hand that the money at MIT is not going to Julia. Okay, there was a few years we had a devops person dedicated to making the JuliaLang organization work (at probably half the salary of FAANG for the love of OSS and MIT prestige), but the vast majority of the money does not go to Julia. For grant money at MIT, more than 50% is siphoned away before it gets to the lab. Then, the vast majority of that leftover money is not for software development but instead for research in methods for parallel computing, scientific machine learning, and applications to XYZ (you can just take a look at my CV or Alan Edelman's if you're curious about what the Julia Lab actually does).

While a lot of this research does trickle back into the Julia ecosystem, it's much less direct than Facebook paying the day salary for a bunch of PyTorch engineers. In fact, it's almost the opposite: most of the people who are paid are graduate students, who may be at risk of graduating if all they do is contribute to Julia instead of getting the required research manuscripts written.

One of the major transformations that has happened to the ecosystem though is that post-1.0 it has gotten some pretty wide industry adoption, along with some major success stories built directly off of the Julia ecosystem. Julia Computing, Pumas-AI, RelationalAI, Invenia, and Beacon Biosystems are just a few companies that are noteworthy of hiring a good number of contributors to the language and package ecosystem. The hiring by these kinds of entities has actually been growing so rapidly that it's a great labor market to be looking for a job in. But most of this hiring frenzy is not driven by MIT.

Thanks for the overview how it actually works.

But then you've to consider that many languages have been created in academic institutions, few have become semi-popular, fewer have entered the industry sector.

Depends in what context, a language created by MIT researchers with the goal of changing the status quo of Python and R, followed by the creation of a foundation to drive its adoption among commercial entities, is quite different than the thousands of programming languages created across the globe for yet another compiler class assignment.

In my opinion the most important role in its adoption was having researchers putting it in application in their field. In a way it was Julia's killer app. Considering it is made for technical computing, this showed that it is useful and actually *used* for its purpose, and not something that some researchers have academic fun with. This is not to say that languages meant for academic purposes are bad but rather that it isn't something which will make a language popular.