skip to content

Top Stories of the Week - 3/15

Weekly Top Stories 3-15-24 - Galaxy Research

This week in the newsletter, we write about Ethereum’s latest upgrade, the rising rates across DeFi, and Jito’s struggles against MEV on Solana.

Subscribe here and receive Galaxy's Weekly Top Stories, and more, directly to your inbox.

Dencun Upgrade Goes Live on Ethereum Mainnet

Ethereum protocol developers successfully executed yet another network-wide upgrade on the world’s second most valuable blockchain on Wednesday, March 13. The upgrade, dubbed Dencun, is the 16th hard fork that developers have activated (excluding upgrades to the Beacon Chain) since the network’s genesis in 2015. Speaking to the success of the Dencun hard fork, Ethereum Foundation Protocol Support Lead Tim Beiko tweeted shortly after the upgrade went live, “Dencun is both the most complex fork we've shipped since the Merge, and tied for ‘most total EIPs in a fork’ with [the Byzantium upgrade]. There were more teams than ever involved in the process, and it somehow all worked out smoothly...! Grateful to work with all of them, onto the next one!”

The most complex code change included in Dencun was EIP 4844, also known as proto-danksharding. EIP 4844 introduces a new type of transaction to Ethereum called “blobs.” Blobs are a cost-effective way for Layer-2 rollups to commit batched user transaction data to Ethereum. Since Dencun was activated, rollup teams including Starknet, Optimism, Base, and Zora have reported major reductions in fees due to their utilization of blobs. According to data on Dune collected by OP Labs, transaction fees on Optimism dropped from roughly $0.50 cents per transaction to less than $0.001 the day of the upgrade. Fees on Optimism have since increased to roughly $0.02 as of March 14, 2024.

Already over 1GB of data has been posted to Ethereum through blobs and this despite the network cost to submit blobs for rollup operators being virtually zero ETH. The cost of blob transactions is expected to increase over time as more rollup operators adopt blobs and as transaction activity on rollups increases utilization of blob space on Ethereum. As stated in prior research notes, Dencun is the first of many upgrades that will enhance Ethereum’s ability to scale user activity through rollups. Another rollup-centered upgrade that developers are tentatively planning to work on now that Dencun is complete is data availability sampling (DAS). DAS will enable nodes to support a higher number of blob transactions per block, for example 64 instead of 6, without increasing the networking and data storage requirements of nodes. For more information about full danksharding, the next iteration of EIP 4844, read this Galaxy Research note.


The activation of Dencun was a resounding success, resulting in no meaningful degradation to network health immediately following the upgrade. Dencun has already resulted in noticeable improvements to rollup transaction fees from the utilization of blobs by rollup operators. With the short-term success of Dencun now in the rear view mirror, Ethereum protocol developers have shifted their focus to preparing for the next major upgrade, Pectra. On Thursday’s All Core Developers (ACD) call, ACDE #183, developers debated what code changes should be prioritized for Pectra. It is clear that like Dencun, there are more code changes than developers have the bandwidth to implement in one upgrade. The need felt by developers for prioritization of a few EIPs among several candidate EIPs for Pectra is particularly acute because developers are aiming to roll out Pectra on mainnet before the end of the year. While they are not likely to achieve this target (as Ethereum developers rarely meet their anticipated timelines for upgrades), the sentiment to scope out Pectra for a hopeful activation in 2024 is fueling debate and controversy over what improvements to Ethereum are more urgent than others.

On ACDE #183, Ethereum Foundation researchers, client developers, wallet developers, and smart contract developers shared impassioned remarks about the relative importance of code changes improving upon a broad range of issues, including: end-user experience, decentralized application (dapp) developer experience, censorship resistance, as well as network stability and security. Despite the long-term vision of scaling Ethereum through rollups, it is clear that Ethereum developers have no intention of halting improvements primarily benefitting end-users and dapp developers transacting directly on Ethereum. Despite efforts to improve the protocol’s flexibility to support rollups, Ethereum’s primary use case remains general purpose computation, not data availability (DA). While not as widely discussed in the Ethereum community, the Dencun upgrade contained code changes to not only reduce fees for rollup users, but also users of decentralized finance (DeFi) apps built on Ethereum.

The main code change in Dencun impacting DeFi users, particularly Uniswap users, is EIP 1153. EIP 1153 introduces two new operations, also called opcodes, TLOAD and TSTORE, that allow smart contracts to utilize stored data and discard the data after every transaction. TLOAD and TSTORE function similarly to existing storage opcodes, SLOAD and SSTORE, but are priced more cheaply because these operations do not update or otherwise make changes to stored data. As estimated by founder of OP Labs, these opcodes will enable dUniswap to save end-users $3mn/year in gas costs based on the exchange’s activities from 2022. While the long-term vision is to migrate all end-users and dapp developers to rollups, Ethereum developers are also working in parallel in the short to medium term (before rollups as a technology are mature and battletested) to reinforce Ethereum’s competitive edge as a general purpose and monolithic blockchain over and above its other Layer-1 competitors. At the same time, Ethereum developers must balance these efforts with pressures to evolve quickly enough to compete with comparatively more performant DA layers wholly focused on servicing rollups, as opposed to end-users directly.

Striking such a fine balance between rebuilding Ethereum’s competitive edge as a DA blockchain while also maintaining a lead over other general purpose blockchains through servicing the demands of its existing userbase, primarily composed of dapp developers and end-users, will not be easy. It will almost certainly open opportunities for competitor networks to siphon users from Ethereum and thus, lead to increased fragmentation in Ethereum’s dapp ecosystem. - Christine Kim

Jito Shutters Solana Mempool

Jito Labs, the leading provider of liquid staking tokens and MEV solutions on Solana, announced the discontinuation of its mempool services. On March 8, the team reported the suspension of their mempool offering, attributing the decision to "negative externalities affecting Solana users."

In proof-of-stake networks, validators take turns assuming the role of "block leader," responsible for selecting transactions to include in the proposed block. On Ethereum, pending transactions accumulate in a mempool, waiting for block inclusion. This opens up the possibility for the block leader to manipulate transaction order or selection to extract additional value, a practice known as MEV.

Solana's architecture diverges from Ethereum's by not incorporating a native mempool. Instead, transactions are streamed directly to current and future block leaders. For example, it is difficult to front-run transactions as they do not sit in a mempool waiting for execution like on Ethereum. Additionally, MEV on Solana exacerbates issues with spam. Because fees on the Solana are currently ~$0.00005, it is positive expected value for MEV actors to spam the network in order to increase their probability of inclusion. 50% of Solana transactions are failed arbitrages.

To mitigate network congestion from spam, Jito introduced an external mempool solution for Solana, effectively adding a 200ms mempool phase within the 400ms block period. Jito's solution aimed to decrease spamming incentives by allowing transaction submitters to secure inclusion through validator tips, mirroring Ethereum's MEV-geth mechanism. The introduction of the mempool, however, also opened the door for more toxic forms of MEV like sandwiching that harm the user experience, ultimately resulting in Jito shuttering the service.

Solana has experienced a surge in MEV related activity as onchain volumes have surged over the past year, consistently ranking as one of the top three chains by DEX volumes. Despite halting the mempool service, Jito maintains its other offerings, such as the block engine and slot auction services. They have not disclosed whether they intend to reintroduce the mempool solution in the future.


MEV continues to be one of the biggest challenges facing blockchains that have transaction ordering, primarily due MEV’s ability to continually morph depending on the characteristics of the underlying blockchain architecture. Jito has spent months working on various fixes, and years prior to that learning from MEV on Ethereum. But for every solution, searchers have figured out a new workaround, resulting in a continuing game of cat and mouse.

The response from the Solana community has been mixed, but largely positive. Leading validators and ecosystem actors have been supportive of Jito’s move, citing the damage sandwiching was doing to the user experience. Those against the decision have highlighted the fact that closing their mempool only opens the door for a less altruistic actor to take Jito’s role. Indeed while sandwiching may be less following the decision, it is still happening.

The decision to shut down the Jito’s private mempool is ultimately an attempt to buy developers more time to explore other solutions. Other ideas include using a request for quote system that routes trades offchain to professional market-makers or trusted RPC endpoints that refund users for MEV extracted from their orders. User and dapp builder education is also important. Sandwich attacks are particularly prevalent on trades involving memecoins, which usually have low liquidity and high slippage. Users that understand how slippage impacts their ability to experience sandwiching, or frontends that automatically adjust fees and slippage based on current market conditions, could help ensure that users who did get sandwiched are aware of the risk they are taking ahead of time.

The developments are a reminder that there is no silver bullet for MEV, only tradeoffs. In this case, Solana is fortunate that Jito is willing to forsake short-term economic gains to identify longer-term more sustainable solutions. This will not always be the case, and it is clear that any solution will require as much a social consensus as a protocol one. - Lucas Tcheyan

Dai rates spike as Maker takes unprecedented action

Maker raises Dai borrowing (& savings) rates to 15%+ to 'prevent Dai demand shock'. Following approval of an Accelerated Proposal submitted on March 8, Maker Protocol raised DAI borrowing rates by ~9-ppt to 15.0%-17.25 across various Maker Vaults on March 10. Other implemented changes as part of the proposal include: (i) increasing the Dai Savings Rate from 5% to 15%, (ii) increasing the Dai Borrow Rate on Spark from 6.7% to 16.0%, (iii) reducing the minimum time needed to implement changes to Maker protocol after an executive vote has passed from 48 hours to 16 hours, and (iv) enhancing USDC deposit and DAI minting throughput in the Price Stability Module (PSM).

Motivations for the proposal were a significant decline in reserves due to volatile market conditions and rapid DAI outflows with users selling DAI for other assets. Specifically, PSM reserves had fallen below $320m at the time the proposal was submitted (down from $400m+ the prior week) while DAI supply stood at $4.3bn (-12% from ~$5.0bn the prior week). In addition, the proposal noted that RWA collateral held by the protocol totaled ~$1.1bn, yet that balance is subject to redemption delays and daily redemption caps. The proposal noted that the immediate changes are intended to be temporary and will revert to normal operations once the current risks are mitigated, and that medium-term actions to strengthen PSM reserves.

Since Maker implemented the changes on March 10, the PSM reserves recovered to nearly $800m while DAI supply expanded by ~$200m (+5%) to $4.5bn as of writing.


MakerDAO's decision to implement these unprecedented rate hikes felt very sudden and reactionary, especially as the changes called for reducing the governance delay module (intended to allow sufficient time for MKR holders to review any scheduled system changes) to 16 hours after a successful vote. That said, Maker did experience a significant decline in its reserve composition over recent weeks that could have left Dai vulnerable to losing its peg if no action had been taken. Since February, the rate environment in crypto has changed dramatically as demand for leverage surged - USDT borrowing rates on Aave, for example, had spiked from 4.4% to 16%. This also drove demand for recently launched USDe (Ethena's delta-neutral stablecoin; covered in our prior newsletter), currently offering 60%+ APY with the favorable basis trade. Maker's borrowing rates, which were heavily dependent on the relatively low T-bill rate, had lagged as a result, leading to significant outflows from the PSM.

Arguments could be made on both sides as to whether Maker's implemented changes were optimal. Looking back, the warning signs that rates would have to increase were certainly there and could have been addressed earlier. It is possible that the rate hikes could have been more gradual. Either way, it's better to be managing these risks with the market on the way up rather than on the way down, which would be a much more concerning position. - Charles Yu

Charts of the Week

Ethereum’s Dencun upgrade was activated at Beacon Chain slot 8,626,176 on March 13, 2024, including EIP-4844 which introduced blobs. Blobs are a cost-efficient way for Layer 2s (L2s) to verify data on Ethereum. In the first 33 hours after EIP-4844 was initiated, L2s have submitted 12,039 blobs. The first blob was submitted at slot 8,626,178. Starknet maintained the greatest share of blobs submitted at

Ethereum Blobs Submitted - Chart

Each blob carries two distinct fees that must be paid by the sender: the blob fee (Consensus Layer) and the Execution Layer fee. The blob fee can be thought of as the fee to create the blob transaction, while the Execution Layer fee is the cost to get the blob transaction executed and validated on Ethereum.

The chart below tracks the total fees paid for blobs and their accompanying Execution Layer transactions. In total, $280k has been paid on Execution Layer transactions and a meager $0.0000059822 has been paid on Blob fees in the 33 hours following the activation of EIP-4844. So, while blob fees are miniscule, the Execution Layer fees are higher. They still represent a healthy discount on the call data fees paid by L2s before the introduction of EIP-4844, which amounted to more $3.22m among the top L2 in the two days leading into Dencun. We will get a clearer picture of the extent to which EIP-4844 is cost efficient as more L2s begin using blobs.

Ethereum Consensus and Execution Layer Fees Spent on Blobs - Chart

Other News

  • Craig Wright is not Satoshi Nakamoto, UK judge rules

  • MicroStrategy plans to sell $500 million in debt

  • Coinbase plans $1B bond sale

  • BNB Chain introduces rollup-as-a-service solution to expand its Layer 2 ecosystem

  • US judge denies motions to dismiss SEC's lawsuit against Gemini Earn program

  • Solana memecoin dogwifhat to grace Las Vegas skyline after $650,000 raise

  • MakerDAO founder announces two new tokens in 'Endgame' launch details

  • Arbitrum eyes $400 million crypto gaming fund with proposal to DAO