One is that the marginal cost of software(1) drives this pattern of winners and losers. The first user of any software costs an enormous amount of money to actually write the software and deliver it to customers. The 100th user costs basically nothing once you have 99 others. And the millionth user (or billionth) user costs basically nothing as well(2). That in turn means that having a billion users is a lot more profitable than having a million users, which means that if you have a billion users you can afford to do things that the million user system can't- e.g. free webmail and a really good free internet browser, just to name two things picked completely at random and not having any particular company in mind.
The other point is explaining your comment about the "last 15 years": tech's dominance (really, growth's dominance) is really an artifact of zero-lower-bounds interest rates from the 2008 financial crisis. If interest rates are zero (for discounted future cash flow computations) then I am indifferent about a dollar today versus a dollar in 2075. So someone who can argue that they have a 5% chance of being worth a trillion dollars in 2075 is worth a lot (0.05 * 1T=50 billion) when interest rates are zero, but if interest rates are high (or even, honestly, normal- like 2-3%) then that money is discounted heavily and the growth story doesn't matter as much because dollars today are worth a lot more than dollars in 2075. So if interest rates are zero, future growth will dominate the stock market (which was why 'tech' did well) but when interest rates are more normal, different companies can dominate the stock market (where the fundamental valuation of a company is, roughly, the expected value of future cash-flows discounted to the present).
1: Delivered by the internet- physical media distorts this a bit and behaves more like normal retail goods.
2: Exceptions for certain points in the growth curve where some key system falls over and needs to be rapidly replaced, e.g. storage or compute or whatever, but outside of those it's very cheap growth. Plumber company growth is limited by the number of trained plumbers you can hire- you can only have 1 plumber make so many house calls in one day- but software just replicates at zero out to infinity (again modulo some key systems which can't handle the load).