Ethereum Layer-2 Networks Hit Record Throughput of 30 Million Gas Per Second

Across all tracked L2 protocols, aggregate processing hit 29.64 million gas per second — a ceiling no one was quoting twelve months back — with Coinbase's Base absorbing roughly two-thirds of that volume. The milestone matters less for the headline and more for what it reveals about where the real bottlenecks in the modular stack have moved.

Reading the 29.64 Mgas/s Number

The single-ecosystem aggregate is a useful figure, but it obscures the more interesting architectural story underneath. When one L2 commands 67% of measured throughput, you're no longer looking at a competitive scaling market — you're looking at a concentration risk that protocol engineers should price into their deployment decisions. Base didn't get there by accident: it ships a sequencer architecture optimized for high-volume consumer flows, and the throughput curve shows it.

The practical question for anyone building on L2 today is whether that scale is sustainable without compromising on the properties that made rollups attractive in the first place — censorship resistance, credible neutrality, and predictable finality. The Yellow.com coverage flags exactly this tension, noting that sustained high volumes will stress the networks' ability to maintain efficient processing. In practice, throughput at this level typically correlates with aggressive batching and compression strategies, which compress the data footprint per transaction but raise the cost of a sequencer failure. If you're anchoring user-facing infrastructure to a single L2, that's the dependency chain worth mapping before you scale.

What Engineers Should Watch Next

Three signals will tell you whether the 29.64 Mgas/s figure represents a new floor or a momentary peak.

First, sequencer decentralization progress on Base. Until the sequencer is meaningfully distributed, the throughput ceiling is also a single-point-of-failure ceiling. Coinbase has signaled a path here, but the timeline remains the variable that matters.

Second, blob utilization on L1. L2 throughput is ultimately bounded by how much data the parent chain can carry via blobs. The current numbers imply the blob fee market is absorbing this load cleanly, but a sustained push toward 30+ Mgas/s will expose whether EIP-4844 capacity is sufficient or whether a follow-up data-availability upgrade is back on the critical path.

Third, competitor rollups closing the gap. If Base's 67% share holds through the next quarter, the conversation shifts from "which L2 wins" to "what is Base becoming" — and that has very different implications for ecosystem neutrality. Conversely, if alternative rollups start absorbing meaningful share via differentiated proving systems or lower-latency finality, the aggregate number stops being a Base story and becomes an L2 maturity story.

The Trade-off Matrix

The honest engineering read is that this milestone is real but conditional. A rollup processing 29.64 Mgas/s across the ecosystem is not the same as one rollup doing it deterministically under adversarial conditions, with censorship resistance preserved and a credible exit path to L1 intact. The throughput number measures the happy path. Production-grade deployments — the kind worth running real value through — require the unhappy paths to hold up too.

For now, the move is straightforward: if your contract architecture assumes L2 throughput will keep climbing, you're betting on continued blob capacity expansion and sequencer reliability improvements. The data says that bet is currently paying off. The open question is how much further the curve bends before the underlying constraints — data availability, sequencer decentralization, proof system maturity — reassert themselves.