Ethereum’s major 2025 upgrade completed: a faster and cheaper mainnet has arrived

On December 4, Ethereum's second major upgrade of the year, Fusaka, was officially activated on the Ethereum mainnet (corresponding to Epoch 411392). This upgrade was implemented on both the consensus and execution layers, enhancing Ethereum's ability to handle large-scale transactions from various Layer 2 networks, increasing the block gas limit to 60 million, and paving the way for subsequent blob parameter adjustments through BPO forks. It is expected to reduce L2 fees by 40% to 60%.

Even before the upgrade was officially completed, it had already received positive attention from some well-known figures in the community. For example, Bitwise Chief Investment Officer Matt Hougan stated on November 23, "The current market correction has caused a lot of information to be overlooked, such as the increasingly impressive value capture capabilities of UNI, ETH, and XRP. I believe the market will soon begin to reprice around the positive impact of the Ethereum Fusaka upgrade, especially if it is delivered as expected on December 3. This is an underestimated catalyst and one of the reasons why Ethereum may lead the current crypto market rebound."

The Fusaka upgrade actually includes a total of 13 EIPs, and introducing all of them one by one may not be very intuitive. So, what changes have occurred after this Ethereum upgrade?

Faster, Cheaper

Let's take a direct look at the current gas levels:

Over the past 7 days, the average Gwei value was about 0.1, while the average Gwei value over the past 30 days was about 0.3. At the same time, compared to the number of transactions in early November, there has not been much change.

Currently, the gas fee for a single USDT transfer on the Ethereum mainnet has dropped to as low as $0.01. I tried such a transfer transaction in a block with about 0.029 Gwei, and it was completed in less than 1 minute.

According to GasFeesNow data, the current gas cost for transferring USDT on the ETH mainnet is only about 0.016% of that on the Tron chain:

Following EIP-7783 from the Pectra upgrade, EIP-7935 in the Fusaka upgrade once again raised the default gas limit, increasing it to 60 million. A higher gas limit means that each block can process more transactions, reducing both block congestion and transaction fees.

According to Chainspect data, the theoretical TPS peak of the Ethereum mainnet has now reached 238.1, whereas in the early days of Ethereum, the theoretical TPS peak was only 15. In 10 years, that's nearly a 16-fold increase—a true "sharpening the sword for ten years" transformation.

PeerDAS, a Major Step Forward for Sharding

"PeerDAS in Fusaka is significant because it is essentially sharding. Ethereum can reach block consensus without any single node having to view even a tiny portion of the data. Moreover, it can resist 51% attacks—it uses client-side probabilistic verification instead of validator voting. Sharding has been Ethereum's dream since 2015, and data availability sampling has been in progress since 2017. Now, we've finally achieved it."

From Vitalik's tweet above, we can see why PeerDAS is the most talked-about feature in the Fusaka upgrade. PeerDAS (EIP-7954) introduces a data availability sampling mechanism, using KZG proofs, allowing nodes to sample a small portion (1/8) of blob data and compare it with data held by peer nodes.

This enables Ethereum to increase the blob capacity of each block without increasing node size, allowing only partial data downloads to verify L2 transaction validity. For L2s, this not only means theoretically 8 times the data throughput, but also lower blob fees and user costs.

(Image source: @jarrodwatts)

However, Vitalik also pointed out that the sharding mechanism in Fusaka is still imperfect in three aspects:

· O(c^2) transactions (where c is the computational power of each node) can be processed at the L2 layer, but not on the Ethereum mainnet. To achieve direct scaling improvements for the mainnet, in addition to the improvements brought by constant factor upgrades such as BAL and ePBS, a mature ZK-EVM is still needed.*

· Proposer/developer bottleneck. Currently, developers need to have all the data and build the entire block. If distributed block construction can be achieved, that would be fantastic.

· We still need a sharded mempool.

Vitalik stated that in the next two years, efforts will continue to improve the PeerDAS mechanism, cautiously scaling it up while ensuring its stability, and using it to scale L2s. Once ZK-EVM matures, it will be applied to the Ethereum mainnet itself to scale the mainnet and continue to improve L1 gas.

Soaring Blob Fees

Blob gas fees have soared by tens of millions of times:

Before the Fusaka upgrade, blob gas was basically free, because there was no minimum fee set for L2s to submit blob data to the mainnet. When mainnet gas rose, L2s would also stop submitting blob data to the mainnet. This meant that the Ethereum mainnet was "working for free" to provide security guarantees for L2s.

Take Lighter as an example. In the past day, they accounted for about 92% of all Rollup L2 TPS, but paid less than $100 in fees to the mainnet. Calculated on a yearly basis, their average daily payment to the mainnet is only about $670.

So it can be said that there was previously a "misalignment of incentives" regarding blob gas fees between the mainnet and L2s. L2s wanted blobs to be as small as possible to reduce user gas fees, while the mainnet wanted blob gas fees to be as high as possible because the mainnet provides security guarantees for L2s.

The EIP-7918 proposal in the Fusaka upgrade linked blob gas fees to the mainnet's gas fees, giving blob gas fees a floor and making them more stable and predictable. Bitwise analysts said that after the Fusaka upgrade, the minimum blob gas fee is roughly the mainnet's base execution fee divided by 16, which will create a more stable income and burn stream for the mainnet.

Of course, the actual increase in income and burn for the mainnet will ultimately depend on the specific growth of each L2.