> With staking, the biggest holders just sit there and collect coins with few operational costs.

Every holder who is staking is rewarded for staking, in proportion to their stake.

> To me it sounds like staking has a much more powerful snowball effect than mining

No, because issuance under proof of stake is much lower than that under proof of work.

The point of validating is to secure the chain, not to get rich. The more validators that are active, the lower the per-validator rewards, reducing compounding effects.

It is much cheaper to run a staking operation but not 0-cost. You still need an internet connection to pay and a bit of energy to run your validator. You may owe taxes on staking rewards, etc... So there is likely to be some selling by validators. Rewards are also much lower, issuance of ETH is cut by 90%, for example.

Also you lock-up capital which has an opportunity cost and must competes with every other investable asset. As the barriers to entry staking are negligible (just acquiring a liquid staking derivative like rETH, for example) it means that everyone willing to stake will likely do so which drops the returns lower, which may push some stakers out as they see better investing opportunities. So staking is likely going to have thin margins in the future and will give a rate of return that is fairly priced given its risk adjusted returns.

Costs are lower, but the rewards are also lower, at about 10% of Ethereum's mining rewards.

Last I saw, ETH mining was about 33% profitable. With mining rewards ten times higher than staking rewards, that means miners take home about three times more on their investment than stakers.

There is some legitimate concern here. On the other hand, PoS looks to be more accessible to hobby miners/stakers because they don't have to convert their coin into ASICs or GPUs.