MegaETH is a real-time Ethereum L2 hyperscaler built for extreme throughput. It features 10ms block times and its architecture is designed to handle transaction volumes far beyond any existing L2.

On January 22, 2026, ahead of mainnet launch, the MegaETH team kicked off a seven-day global stress test: a coordinated push to see if the chain could hold up under real load. During that test, 10ms blocks were consistently produced and at peak, the network processed 35,000 transactions per second (TPS). 10.7 billion transactions were processed over the seven days, putting MegaETH's real-time claims to the test and passing with ease.

High TPS means high data availability demand. Every transaction needs to be published somewhere so the network can verify it. During the stress test, MegaETH posted all of this data to EigenDA, totaling over 1,000 GiB in just seven days.

A useful way to understand DA efficiency is to look at how many transactions fit into each MiB of data posted. During the stress test, MegaETH achieved around 10,000 transactions per MiB, considerably higher than the ~4,000 TX/MiB seen under normal load today. This makes sense: stress test transactions tend to be simple and uniform, which compresses well and packs tightly. Under normal load, transactions are more varied and complex, increasing the average data footprint per transaction.

There's also a structural reason efficiency improves under load: each block carries fixed overhead (headers, metadata) regardless of how many transactions it contains. When blocks are full, that fixed cost is spread across more transactions, lowering the per-transaction DA cost. The same principle applies to compression: denser, more repetitive data compresses better.

The stress test makes one thing clear: MegaETH needs a DA layer built for hyperscale. Let's look at the two candidates, Ethereum blobs and EigenDA, and see how they stack up.

Ethereum blobs (EIP-4844) were a major step forward for L2 DA costs, but they were never designed for this scale. The entire Ethereum blob layer, shared across every rollup, currently maxes out at around 0.145 MiB/s. Assuming a conservative 4,000 TX/MiB, that translates to a TPS ceiling of roughly 580 TPS for all rollups combined. Even if MegaETH were the only L2 posting to Ethereum blobs, that capacity would still fall more than 10x short of what the stress test required. At peak, MegaETH was pushing around 2 MiB/s to EigenDA.

EigenDA is a purpose-built DA layer on EigenLayer, designed to scale far beyond what Ethereum blobs can offer. At its current capacity of 100 MiB/s, EigenDA can already support up to 400,000 TPS for MegaETH (assuming a 4,000 TX/MiB). As EigenDA scales toward 1 GiB/s, the numbers become staggering. Millions of TPS become theoretically possible.

The charts below compare Ethereum blob TPS capacity against EigenDA's, showing how much room MegaETH has to grow.