Pectra activated on 7 May 2025 at epoch 364032, 10:05:11 UTC — exactly one year ago today. Seven months later, on 3 December 2025 at slot 13,164,544, Fusaka followed. Together these two upgrades did more to reshape Ethereum than anything since the Merge, and the network you're looking at in May 2026 is fundamentally different from the one that ended 2024.
This is Part 1 of a four-part series unpacking Ethereum's post-Pectra, post-Fusaka reality. We'll start wide: what actually changed, what it means for ETH the asset versus ETH the network, and what's queued next. Later parts go deep on PeerDAS, validator economics after MaxEB consolidation, and the road to Glamsterdam.
Pectra: the kitchen-sink upgrade
Pectra shipped 11 EIPs in a single hard fork — the largest EIP count in Ethereum's history. The headline acts:
- EIP-7251 raised the validator MaxEB (maximum effective balance) from 32 ETH to 2048 ETH, allowing solo stakers and large operators to consolidate dozens of 32-ETH validators into single beefier ones. This was the single biggest operational change to the consensus layer since the Beacon Chain launched.
- EIP-7702 introduced delegated EOAs — externally-owned accounts can now temporarily adopt smart-contract code for a single transaction. It's account abstraction without forcing users to migrate to a new account type.
- EIP-7691 doubled the blob target and max from 3/6 to 6/9 per block, immediately giving rollups roughly 2x cheaper data availability.
- EIP-2537 added a BLS12-381 precompile, which made aggregated signature verification dramatically cheaper — important for rollups, bridges, and zk systems.
- EIP-6110 and EIP-7002 moved validator deposits and withdrawal initiations onto the execution layer, eliminating multi-day deposit queues and giving stakers protocol-level exit triggers.
- EIP-7549, EIP-7623, EIP-7685, EIP-7840 rounded out the fork with attestation reform, calldata repricing, a generic execution-layer requests pipeline, and blob-schedule configuration.
If you want to understand why so many of these matter, it helps to know how Ethereum's account model and EVM actually work — Pectra touched the validator account model, the EOA semantics, and the precompile set in one shot.
Fusaka: PeerDAS and the data-availability pivot
If Pectra was the kitchen sink, Fusaka was the surgical strike. It activated at slot 13,164,544 (epoch 411,392), 21:49:11 UTC on 3 December 2025, and its centerpiece was a single mechanism: EIP-7594, PeerDAS.
PeerDAS (Peer Data Availability Sampling) lets validators verify that blob data was published to the network without each one downloading every blob. Each validator samples a small subset; the math guarantees that if enough honest samples succeed, the data is recoverable. This is what unlocks meaningful blob count increases without forcing every node to scale linearly with rollup demand.
Fusaka also shipped:
- EIP-7825 — a 16,777,216-gas per-transaction cap, putting a hard ceiling on any single transaction's compute footprint.
- EIP-7935 — raised the default block gas limit to 60M.
- EIP-7918 — a blob base-fee floor, so blob pricing can't collapse to near-zero and starve the fee market.
- EIP-7951 — a secp256r1 precompile, enabling cheap verification of passkey/WebAuthn signatures on-chain.
The combined effect: Ethereum L1 became measurably better at being a data-availability layer for rollups, and meaningfully better at consumer wallet flows.
Blob Parameter-Only forks: the new cadence
Fusaka introduced something genuinely new to Ethereum's release process — BPOs (Blob Parameter-Only forks). Instead of waiting for the next major hard fork to raise blob counts, the network now ships small, pre-scheduled bumps:
- BPO1 on 9 December 2025 raised blob target/max to 10/15.
- BPO2 on 7 January 2026 raised it again to 14/21.
That's roughly a 2.3x increase in blob throughput in five weeks. Validators also raised the L1 gas limit from 45M to 60M on 25 November 2025, with social-layer discussion already pointing toward 150M targets later in 2026 as PeerDAS performance data accumulates.
This matters because rollups are now firmly Ethereum's scaling story. Arbitrum One sits at ~$15.7B TVL, Base at ~$12.2B, and combined L2 TVL is in the $30-35B range as of Q1 2026. Cheaper, more abundant blobs translate directly into cheaper rollup transactions, which is the only scaling lever the rollup-centric roadmap actually pulls.
ETH the asset: ultrasound-money, conditionally
Here's the part that's harder to spin. ETH supply at the start of 2026 sits between 120.7M and 121.5M, with annualized net inflation in the 0.2-0.3% range. ETH is mildly inflationary in 2026 — not deflationary.
The "ultrasound money" thesis was always conditional: ETH is deflationary only when burn (from EIP-1559 base fees) exceeds issuance (to validators). Post-Dencun, post-Pectra, post-Fusaka, a growing share of activity has migrated to L2s, which pay for blobs but burn far less ETH per unit of user activity than equivalent L1 transactions did. Issuance, meanwhile, is roughly steady because the validator set is large.
Result: about 35.8M ETH staked (roughly 28.9% of supply) earning ~3.3% APY, paid for by mild dilution of non-stakers. This isn't a crisis — 0.2-0.3% inflation is still tighter than almost any fiat — but it does mean the marketing line has to be retired or rewritten. Honest framing: ETH is sound money with a productive yield, not a hyper-deflationary asset.
If you want the long arc on how supply got here, the post-Merge supply history is the right next read.
Validators after MaxEB
EIP-7251's MaxEB increase from 32 to 2048 ETH triggered the largest restructuring of the validator set in Ethereum's history. Solo stakers consolidated multiple validators into one. Large operators (Lido, Coinbase, Kiln, Figment) consolidated thousands. The total active validator count dropped sharply through mid-2025 even as the total staked ETH kept climbing.
This is good for the protocol — fewer validators means less peer-to-peer messaging overhead, faster attestation aggregation, and a more sustainable path to single-slot finality. It's complicated for the staking-as-a-service market because consolidation reshapes the unit economics of node operations.
We cover the full validator and LST landscape separately — it's a topic that deserves its own piece.
What's next: Glamsterdam, and the long tail
The next hard fork is Glamsterdam, targeted for H1 2026. Two EIPs are confirmed for inclusion:
- EIP-7732 — ePBS (enshrined Proposer-Builder Separation), pulling MEV-Boost-style block production into the protocol itself.
- EIP-7928 — Block-Level Access Lists (BAL), enabling parallel transaction execution and more efficient state access.
EIP-7805 (FOCIL) — Fork-Choice-enforced Inclusion Lists — is a Glamsterdam candidate still in Draft. It would push back against censorship by mandating that block proposers include certain transactions. Important to be precise: FOCIL is not in Fusaka. It's a 2026 fight, not a 2025 ship.
Beyond Glamsterdam sits the long tail: single-slot finality (SSF), Verkle Trees (for stateless clients), and EOF (EVM Object Format). None have shipped. All have been on the roadmap for years. SSF in particular is the holy grail — a 12-second finality window instead of the current ~12.8 minutes — but it's a multi-fork project, not a single EIP.
So what does this all mean?
Vitalik's rollup-centric roadmap, articulated back in 2020, has been executed almost exactly as written. The L1 is no longer trying to scale execution. It's trying to be the fastest, cheapest, most credibly-neutral data-availability and settlement layer it can possibly be, and let rollups handle execution.
Pectra and Fusaka were the upgrades that made that pivot real. ETH the network is healthier than it has ever been. ETH the asset is mildly inflationary, productively staked, and structurally tied to L2 throughput rather than L1 burn.
If you hold ETH, you can self-custody it in Zelcore alongside the rest of your portfolio — wallet flows didn't change with these upgrades, but the asset they hold absolutely did.
Next up — Part 2: PeerDAS Deep Dive. We'll unpack exactly how data-availability sampling works under the hood, why it's the linchpin of every blob-count increase to come, and what it means for the difference between full nodes and sampling nodes in a post-Fusaka world.
