How Pectra and Fusaka Reset Ethereum's Economics
Two protocol upgrades reset how Ethereum prices rollup data and how validators are structured. The mechanics matter more than the headlines suggest.
Simon Belanger 5 min read
Protocol upgrades on Ethereum are usually reported as a name and a date, then forgotten. The two most recent ones deserve more attention than that, because between them they changed two things that determine what the network actually costs to use and what it pays to secure: how rollup data is priced, and how validators are structured. Pectra arrived in May 2025 and Fusaka in December 2025, and the follow-on parameter changes ran into January 2026. The effects are still working through the system.
What Pectra actually changed
Pectra activated on Ethereum mainnet on May 7, 2025 at epoch 364032, bundling eleven separate improvement proposals. Three of them carry most of the economic weight. EIP-7251 raised the maximum effective balance for a validator from 32 ETH to 2,048 ETH, meaning a stake that previously required 64 separate validator keys could be held in one. EIP-6110 moved validator deposits on chain and cut the activation delay from roughly nine hours to roughly thirteen minutes. EIP-7002 gave withdrawal credential holders the ability to force a validator exit from the execution layer, which matters when the party holding the keys and the party holding the economic exposure are not the same entity.
The fourth item worth naming is EIP-7702, the account abstraction feature that lets an ordinary externally owned account temporarily execute contract code for the duration of a transaction. It moved fast: The Block reported more than 11,000 authorizations in the first week. Adoption has not been uniformly benign, though. A study covering 3.66 million authorization transactions across seven chains found a majority linked to attacker-controlled contracts, mostly automated sweeping of already-compromised keys. The feature works as designed; the design simply also makes draining a compromised account more efficient.
Pectra also doubled blob capacity through EIP-7691, moving the per-block target from three to six and the maximum from six to nine. That was an interim measure. The real change came later.
Fusaka and the blob parameter dial
Fusaka activated at slot 13,164,544 on December 3, 2025, and its headline feature was PeerDAS. Before it, every node that wanted to verify blob data had to download all of it, which put a hard ceiling on how much blobspace the network could offer without pricing out home operators. PeerDAS replaces that with data availability sampling, allowing each full node to hold roughly one-eighth of the data while still verifying that the whole set is available. The ceiling moved.
Fusaka also introduced Blob Parameter Only forks, minimal configuration-only upgrades that adjust blob target and maximum without a full fork cycle. Two ran almost immediately. BPO1 on December 9, 2025 lifted the target to 10 and the maximum to 15. BPO2 on January 7, 2026 lifted them again to 14 and 21, roughly 2.3 times the pre-Fusaka figure, inside five weeks. The Ethereum Foundation’s January 2026 checkpoint noted that developers were holding further increases rather than pushing on schedule, reasoning that a third BPO is not urgent until rollup usage consumes the space already created. Capacity is now a dial that can be turned in response to demand rather than a parameter locked in for a year at a time.
Why cheaper blobspace is not free blobspace
The part of Fusaka that gets least attention may be the one with the longest economic tail. In the period after Dencun, blob base fees frequently collapsed to a single wei, because the fee adjustment mechanism ignored the fact that execution costs dwarfed blob costs. A ten percent move in a fee that rounds to nothing changes nothing. Rollups paid almost zero for data that nodes still had to cryptographically verify.
EIP-7918 fixes that by establishing a reserve price for blobs pegged to L1 execution costs. When the execution gas equivalent of posting a blob exceeds what the blob fee would collect, the fee stops falling and begins responding to demand again, settling at roughly one sixteenth of execution costs. Rollup fees can still fall as capacity rises, but the era in which data availability was effectively subsidized to nothing is structurally over. For anyone modelling rollup margins or ETH burn, that is the assumption that changed.
Validator economics after the 2,048 ETH ceiling
The consolidation permitted by EIP-7251 has visibly reshaped the validator set. Figures reported in early August 2026 put staked ETH at 41.41 million, roughly 34 percent of circulating supply and an all-time high, spread across approximately 893,000 active validators. At the old 32 ETH ceiling that quantity of stake would have required close to 1.3 million validators. The gap is consolidation, and it directly cuts operator costs: the same stake now runs on far fewer keys, fewer machines, and fewer attestation messages propagating across the network.
The offsetting pressure is yield. Ethereum’s issuance is spread across whatever stake shows up, so a record staking ratio mechanically compresses the base return. The same August 2026 reporting put the seven-day staking APR at 2.66 percent, down from a peak above 5 percent in 2023. Consolidation improves the cost side of a staking operation at precisely the moment the revenue side is thinning, which is why the efficiency gains from EIP-7251 matter more to professional operators than the headline number suggests.
Glamsterdam and what is still unsettled
The next upgrade, Glamsterdam, is built around two proposals: EIP-7732, which enshrines proposer-builder separation in the protocol itself, and EIP-7928, which adds block-level access lists to enable parallel transaction execution. Developers narrowed roughly fifty candidate features to seventeen and are adding them to devnets incrementally, with the option to remove any that cause trouble. As of the Foundation’s own January 2026 update, a firm timeline was pending the first stable ePBS devnet, and anyone quoting a confident mainnet date is guessing.
What it means for a Canadian investor
For a Canadian holding ETH through a domestic exchange-traded product rather than a wallet, the validator side of this is not abstract. Canadian issuers were early to staking: 3iQ launched staking in its Ether funds as North America’s first staking exchange-traded products, and CI Global Asset Management moved to add a staking strategy to its Galaxy Ethereum ETF. Staking rewards accrue to net asset value, so the compression in base staking yield described above flows through to those products directly, net of fees. The consolidation efficiencies, by contrast, accrue to the custodian and operator rather than the unitholder.
On the transaction side, the takeaway is that Layer 2 costs are now governed by a supply dial developers can turn and a fee floor they deliberately installed. Cheaper is likely; free is not on the roadmap. And for anyone using a wallet that has adopted EIP-7702 delegation, the same feature that enables gas sponsorship and batched transactions also makes a leaked key more dangerous than it was before, which is an argument for hardware custody rather than against the upgrade.