On top of that, you have to factor in luck: block proposals are randomly assigned to validators, and they include a random amount of tips depending on network activity. You could propose a block that nets you 0.01 ETH in tips, or a block with 50 ETH in tips.
On average, I believe the post-Merge APY has been estimated to be ≈6 % (non-cumulative).
Hardware can be anything. Even a RasPi 4 with 8 GB of memory can be made to work. Any old x64 machine with two or more cores works. 16 GB of memory is recommended, as is a 2 TB SSD to minimize down-time. Storage requirements are likely to go down in the medium-term, as e.g. state expiry get implemented.
I’d recommend an Intel NUC. They use laptop CPUs, so electricity consumption is around 5–20 watts. For networking, you’ll need around 5 Mb/s up and down after the initial sync.
> You should run mev-boost to earn a fair share of the MEV extracted in the blocks you propose. Connecting your consensus client to mev-boost allows you to get full blocks from a network of block builders optimized for MEV extraction. This can increase validator rewards by 75.3%, or give an APR of 12.86% rather than a non-MEV APR of 7.35% from staking eth.
I like the idea of a non-PoW blockchain, and I don't mind taking some risks, but before I invest $60k in becoming a validator, I'd like to know what the potential payback might be.
[1] https://ethereum.org/en/staking/solo/