What Happened to Ethereum?
What Happened to Ethereum?
For much of 2021–2024, the public narrative around Ethereum felt oddly contradictory: the ecosystem was exploding with activity, yet users complained about fragmentation, bridges, and a growing sense that “everything important moved to L2.” In early 2026, that narrative shifted again—sparked by Vitalik Buterin’s argument that the original idea of L2s as “Ethereum-branded sharding” no longer matches reality, because Ethereum L1 is scaling meaningfully on its own while many L2s remain far from Stage 2 decentralization (and interoperability is still lagging). A mainstream recap is available in this report from Decrypt: “We Need a New Path”: Vitalik Buterin Rips Up L2-Focused Roadmap.
So what actually happened to Ethereum—and what should users do next?
This article explains the shift in plain terms: why Ethereum leaned into rollups, why L1 scaling is accelerating again, why “Stage 2” matters, and what all of this means for security, UX, and holding ETH in 2026.
1) Ethereum didn’t “lose”—it modularized
Ethereum’s post-merge roadmap emphasized a modular architecture: keep Ethereum as the most secure settlement and data availability layer, while letting rollups scale execution.
The turning point for rollup economics was Proto-Danksharding (EIP-4844), which introduced blob-carrying transactions—a cheaper, temporary data lane designed primarily for rollups. The spec is here: EIP-4844: Shard Blob Transactions. In practice, blobs reduced the cost of posting rollup data to L1 and enabled much lower L2 fees, helping Ethereum remain competitive without turning L1 into a hardware arms race.
Vitalik’s longer 2025 essay, Scaling Ethereum L1 and L2s in 2025 and beyond, framed this approach as a “stay the course” strategy: L2s scale user transactions, while L1 scales blob throughput and selectively scales execution for what must remain on L1.
2) The “L1 comeback” is real: gas limits and blob throughput are rising
The key update in 2025–2026 is that Ethereum is not only scaling for rollups—it is also scaling itself more aggressively than many expected.
L1 execution scaling (gas limit)
In August 2025, the Ethereum Foundation explicitly set a near-term direction: accelerate L1 execution scaling, with a stated aim of reaching 100M gas over time (with incremental increases and heavy client hardening). See: Protocol Update 001 – Scale L1.
Separately, Ethereum core contributors proposed standardizing a higher default L1 gas limit in clients for the Fusaka era. See: EIP-7935: Set default gas limit to 60M.
Blob scaling (data availability)
On the data side, Ethereum’s plan is even more direct: increase blob throughput so rollups can reliably offer cheaper fees at scale.
- The EF’s blob-scaling strategy is summarized in: Protocol Update 002 – Scale Blobs.
- The flagship technology is PeerDAS (Peer Data Availability Sampling), specified as: EIP-7594: PeerDAS.
- The network upgrade that brought these ideas into a concrete timeline is Fusaka, with key details in: Fusaka Mainnet Announcement.
If you step back, the message is simple: Ethereum is treating blob capacity and L1 performance as first-class deliverables, not “nice-to-haves.”
3) Why Vitalik says “branded sharding” no longer fits
The original “L2s as shards” mental model worked best under two assumptions:
- L1 will remain relatively scarce for execution, so most activity must move to L2.
- L2s will mature into Stage 2 rollups, with minimal trusted control and robust, standardized interop—so they feel like shards of one system.
In 2026, both assumptions look shaky:
- L1 is scaling faster than expected (gas limits and client performance are moving).
- Many L2s are still not “no training wheels” systems, and some may never want to be—due to product constraints, governance preferences, or compliance needs (as summarized in the Feb 3, 2026 coverage above: Decrypt report).
This is where the Stages framework becomes crucial: it helps users separate marketing (“we’re an Ethereum L2”) from trust minimization reality.
A clear explainer is maintained by L2BEAT here: The Stages Framework. In short:
- Stage 0: significant training wheels / governance control
- Stage 1: limited training wheels with stronger proof-based guarantees
- Stage 2: “no training wheels” (code is meaningfully in charge)
Vitalik’s point is not “L2s are bad.” It’s that Ethereum should stop pretending every L2 is a de facto shard of Ethereum with the same social contract—and instead acknowledge a spectrum of security models.
4) What Ethereum is becoming: a spectrum, not a single scaling doctrine
A more accurate 2026 model looks like this:
- Ethereum L1: scaling blobs aggressively (for rollups) and scaling execution enough to handle settlement, exits, proofs, high-value DeFi, and critical infrastructure.
- Rollups: still essential, but their “job description” expands beyond “cheap fees.” The strongest rollups will differentiate on trust minimization (Stage 2), UX, interoperability, specialization, and latency—not just marketing.
- Users: must increasingly make informed choices about tradeoffs (control surfaces, upgrade keys, withdrawal guarantees), instead of assuming “L2 = Ethereum security.”
For a canonical overview of the long-term direction, Ethereum’s official guide is still the best starting point: Scaling Ethereum (ethereum.org).
5) The user checklist: what matters when you move between L1 and L2 in 2026
As costs drop and options multiply, the main risk for users is no longer just “high gas.” It’s confusing trust assumptions.
Here is a practical checklist before you park meaningful funds on any L2 or bridge:
A) Verify the rollup’s decentralization stage
Use an independent dashboard, not a marketing page. Start here: L2BEAT Stages.
B) Understand who can pause, upgrade, or override
If a small group can rapidly upgrade contracts, pause withdrawals, or change proof parameters, your real security model includes that group (and their operational security).
C) Treat bridges as security boundaries
Even in a rollup-centric world, bridge design and upgrade authority remain common failure points. Prefer the most battle-tested, transparent, and minimally trusted routes—and avoid moving large balances “just because fees are cheap.”
D) Keep signing risk under control
Lower fees can increase signing frequency (more approvals, more interactions, more chances to sign something malicious). Good habits matter:
- minimize token approvals
- separate “hot spending” from “cold storage”
- verify addresses and transaction intent carefully
This is also where a hardware wallet can materially help. A device like OneKey keeps private keys offline and forces critical approvals onto a separate confirmation surface—useful when you’re navigating multiple networks, dApps, and bridges, and when phishing is often more dangerous than protocol risk.
6) So… what happened to Ethereum?
Ethereum didn’t stall. It executed a multi-year modular scaling plan (blobs + rollups), and now it’s entering a phase where L1 scaling is accelerating enough to change the social narrative:
- L1 is becoming more capable (execution scaling + client hardening).
- Blob throughput is on a deliberate growth track (PeerDAS and beyond).
- L2s are still central—but must earn the “Ethereum shard” vibe through decentralization and interoperability, not branding.
In other words: Ethereum is not abandoning L2s. It’s abandoning the assumption that all L2s will become uniform, shard-like extensions of Ethereum on a predictable schedule. And for users, that’s a healthy clarification—because it replaces slogans with verifiable properties.
