More ether is staked than at any point in the network's history. Fewer distinct entities are doing the staking. Both statements are true, and protocol researchers increasingly treat the second as the more important one.
Public attribution data maps validator keys to operators with useful but imperfect accuracy. Using the most conservative published mapping, the share of validators run by the ten largest operators has risen steadily over the past year.
The concern is not censorship in the abstract. It is that a small set of operators sharing the same client software, the same cloud region and the same relay set turns an independent failure into a correlated one.
Proposed mitigations under discussion include distributed validator technology, execution-layer changes that reduce the advantage of scale, and reward curve adjustments. None are scheduled for the next upgrade.
Methodology: figures in this article come from publicly published beacon chain data and open-source operator attribution sets. Attribution is probabilistic; where estimates diverge, the lower concentration figure is cited.
