Or don't handle PII, be reasonable with the technical scope of your platform, and have a smaller staff. Bitcoin, for example, was coded by a single person (or at most a tiny group). Even today there are only about 30 people actively writing code for it. You don't need a big team if you link to standard libraries and don't bloat up with gratuitous features.
Platforms that grow too big will suffer the fate of a hot nightclub that makes a splash (and a ton of money) but can't sustain the excitement and goes out of business. Eventually the format burns out and the most desirable people find a new venue, or social competition gets so hot that gangsters start showing up and violence breaks out. Social media will then fall into the pattern of social everything-else. Trendy venues come and go, and smaller venues must find a sustainable niche (craft beer, decent pool tables, or what have you). Slashdot, after all, is still here. I'm glad they didn't try to get too big back in their heyday.
One point I have not seen so far is a rebuttal of the notion that Bitcoin mining is purely wasteful. The resources spent to mine Bitcoin does, in fact, buy something very valuable: it buys security. An ideal world would not need security, and everything spent on security is, on a theoretical level, wasteful. All cryptography (just make it a rule not to snoop), every lock (don't open doors when not allowed), every weapon (just agree not to fight), every fence (don't walk that way), and so on. Enough people believe in the value of trustless digital money to give Bitcoin billions in market cap, so spending something to secure it would make perfect sense in any other context.
I don't think it is a coincidence that New York, home of a cartel that would lose immensely in a Bitcoinized world, and the pioneer of oppressive crypto regulation, would have a hostile attitude to a mining operation, and I suspect that the environmentalist types behind this hit-piece are (perhaps unwitting) patsies for well-connected bankers and their crony political insiders.
I would also add that expansive copyright law, in particular the ban on non-licensed multi-platform client software (ruled contributory infringement in Facebook v Power Ventures) entrenches incumbent platforms with government power. I am old enough to remember when
By that reasoning 200k total for the whole thing seems reasonable, but figures in the millions are simply impossible given limitations of land area and transport capacity. Estimates of 30k (assuming, charitably, they are honest) would have to be based on photos taken early in the morning.
When in a similar situation, AMD spun off its manufacturing and became a fabless company. That, at least, is consistent with the general trend of stratification in the industry, and from what I have seen from the inside of Intel, the fab is 352 organizational layers away from anyone doing design anyhow, so what value is there in such vertical integration anymore?
Intel management seems to want to have it both ways here: getting access to capacity at the latest nodes, without having to manage the risk of building it out. The result to me looks like a half-measure more likely to fail than either committing to vertical integration or going fabless.
You need UADE. It gets faithful results the way SID players do...by emulating a headless Amiga. The source code contains dis-assembled Amiga machine code for the various tracker players. Some of my favorite mod's won't play correctly on anything else (except my real Amiga, of course, but she is getting rather rickety in her old age).
"It is hard to overstate the debt that we owe to men and women of genius." -- Robert G. Ingersoll