Incorrect. As GP pointed out, tech salaries were relatively in line with other mid-level professional roles until the late 2000s, well after software and networked automation had transformed many industries. This was because, at the time, tech companies had to actually have a business plan that sought and gained profits relatively early in the company's life.
What changed - what allowed college dropouts to become billionaires - was first the pre-GFC bubble market (making their net worths essentially inflated). Afterwards, the endless rounds of money-printing used to escape a depression and then sustain growth drove high tech salaries themselves. Without this liquidity, startups would not have had access to VC money and would have had to turn a profit (not just attract further rounds or a successful IPO) or perish within the first few years of existence, kneecapping competition for workers.
This is also why tech salaries in Europe and Asia have stayed relatively sane. Save Japan, they didn't run massive QE regimes and so did not see the same VC environment develop. Why do you think tech companies were the first to see large layoffs after the FFR started rising? It wasn't just that they'd overhired during COVID; that personnel was genuinely needed for the projects they'd had planned. High interest rates screw mightily with their business model, particularly labor pay. You say that demand outpaces the supply of workers, so how could we see layoffs at all? Simple: it's the only alternative to salaries falling when the labor budget is shrunk (through a complex chain of knock-on effects that turn monetary policy into fiscal consequences) by the money printer getting turned off.
As for why tech, and not other industries: in software, networking, etc., it's easy to obfuscate costs and value, particularly when it's unclear what can or has been automated, what a service is worth to a customer, what that customer is actually paying with and for, etc. Arbitrage opportunities abound, and all you have to do is set up shop and choose a growth metric that you can sell the market.