How much BTC/ETH could Google mine if Goog put all resources into mining?

You got a lot of misconceptions about how blockchain works. First of all every block is mined at fixed time(for BTC its 10 minutes). If recent block is mined faster or slower then 10 minutes difficulty will change and block time will always be closer and closer to 10 minutes. Miners get their payment in bitcoins for mining blocks and only first one network who is lucky to mine it gets that block reward. In "Google case" google may get all of blocks and using more and more servers won't get any more money for google. Second: Premium(profit) of mining blocks consists of two parts: block reward + transaction fee. First is hard-coded into algorithms and it started from 50 BTC per block and goes down to 1 and after to 0 by time. No matter how difficult block was reward is hardcoded by number of block. Third: Blockchain is very slow network it allows only 1 block per 10 minutes and block size is around 1 MB. This megabyte is mostly consists of transactions and only 3600 transactions can get into one block(roughly its 6 transactions per second but really its 3-4). Because of it not all transitions get into block and block miner can choose which one to add, so block miner chooses one's that have higher transaction fees. Transaction is an electronic message that consists of: sender address, target address, sum of bitcoins, transaction fee and signature of operation. When transaction is added into block its verified and miner get transaction founds sender included in transaction and guaranteed block reward(1 btc for entire block). So: No matter how much servers are mining - transaction fee is choosen by sender to be in some block(to be in first block today you need to pay arround $25 in btc). Third: Transaction's signature is most important part for security. By choosen algorithm it will take millions of years to break it. So even if someone is successful on 51% attack he won't get ANY founds from bitcoin wallets. 51% attack also won't google change the rules of system so doing it is just pointless. Things are a bit more complicated than I described(its already hard-readable) and etherium like networks are deferent( BTC is proof-of-stake ETC is proof-of-work). Shortly saying putting more servers on etherium will make etherium operations even cheaper.

I quick search for "how many servers google" pops up 900,000 servers. My guess they would be CPU based.

The current state of the art in Bitcoin mining is to use ASICs designed for Bitcoin mining. ASICS have speed capable of reaching 60GH/S (60 billion hashes per second) or more, which is over 1,200 times faster than the average computer.

So divide your 900,000 by 1200 and we get 750.

So Googles entire CPU based servers pointed at Bitcoin would generate the equivelant of only 750 ASICs.

I imagine some of the larger Bitcoin mining farms have many more ASICs than this.

I see there's a sibling "ask HN" question about using AWS to mine BTC.

this suggests an obvious question:

how many magical HN internet points would this "ask HN" thread get if Google put all resources into upvoting this thread?

and if AWS dedicated all resources to upvoting the sibling "ask HN" about using AWS to mine BTC, which "ask HN" thread would win the most magical HN internet points?

Wouldn't they lose more money from not running all their services than they could possibly mine?

A coin is worthless if you own all of it. It's only valuable if everyone owns some of it.