From the 1970s to about 2000, venture capital firms as a class were profitable. Since 2000, venture capital as a class has been a lose. What keeps this going is that each VC thinks they're better than average. Most of them are wrong.
Venture capital in Silicon Valley used to be about finding someone who had a good technical idea, and getting them enough money to make a working prototype or a small production run. New technology and intellectual property were the key. That was a good business.
In the first dot-com boom, this changed. Technology wasn't the issue. It became about buying market share to achieve a "first mover advantage". That resulted in a focus on growth and a race to out-spend the competition. There were winners and losers, but in the end, mostly losers.
In this, the tail end of the second dot-com boom, we see the same pattern. The big difference this time is that nobody is going public. There's just round after round of private capital. The extreme case is Uber, with a valuation greater than General Motors while still losing money at a huge rate.
This is fueled by low interest rates. There's so much capital sloshing around looking for yields that too much money is being funneled into marginal companies. That's why there are no IPOs; it's cheaper to borrow.