Because if it's expected value, I doubt "entrepreneur" is high reward compared to FANGs "moderate reward".
EV = Probability * Outcome
Probability is very low of an IPO outcome - but EV is probably still higher than FANG work.
Suppose, 10^-9 odds of a startup making 10 billion. Let's assume the other terms are negligible.
E_startup = ... + 10^-9 * 10^11 = 100$ E_faang = 10^-1 * 10^5 + 10-2 * 10^6 + ... >= 10^4
Intuitively those numbers feel wrong. theres 7bn people on this planet - and only a very tiny proportion might become tech entrepreneurs. So I think your "1 in a billion" chance of being successful is a bit high. Also, who said $10bn was successful? It is just 1 outcome of many...
Total market value of unicorns ~$2T (~600 unicorns). So average unicorn valuation ~3.3B USD.
Let's say you're the founder and your [ultimately diluted] share of the company is 10%. You have a 10Y runway. Your EV in startup case is 330M (10% of 3.3B) * 1% chance of success = $3.3M or $330K USD per year. This is only counting the extreme (unicorn cases) and will obviously vary with equity percentages.
https://news.crunchbase.com/news/private-unicorn-board-now-a...
$3.3m in NW is 100% beatable in FANG.
It's a tricky problem in the general case because the skills that would allow you to capture a high compensation in FAANG probably correlate with the skills that would lead to a successful startup (not perfectly, but somewhat), so the better your options in FAANG the better your potential outcome via startup (with much higher variance).
FANG: low variance + moderate reward
vs
ENTREPRENEUR: high variance + high reward
The Kelly Criterion can probably be applied here on how much to "invest" in each opportunity. Ultimately it matters how often these opportunities come up.