Lightspeed
July 5, 2025

How To Improve Solana's Market Structure | Eugene Chen

Eugene Chen, co-founder of Ellipsis Labs, breaks down the thorny issues in Solana’s market structure, explaining why his team is building the Atlas L2 and how crypto’s core incentives need a fundamental rethink to compete with traditional finance.

The L2 Solution: Why Atlas is Necessary

  • "On a centralized sequencer layer 2, you can control all of the transaction ordering... Whereas on the layer one, just like some random guy around the world is running your matching engine for 1.6 seconds... It turns out it's very difficult to run an efficient exchange in that type of environment."
  • Solana’s L1 design, with its rotating "proposer monopoly," makes it nearly impossible to run a consistently efficient high-frequency exchange. An unpredictable validator can reorder or interfere with transactions, degrading market quality.
  • Ellipsis Labs is building Atlas, an opinionated Layer 2, as a direct solution. It's not a general-purpose chain; it's a specialized environment designed to be the best possible home for exchange applications like its flagship order book, Phoenix.
  • By controlling the sequencer, Atlas can implement benevolent transaction ordering—for instance, prioritizing critical oracle updates or market maker orders over predatory high-frequency trades to ensure deeper, more stable liquidity.

"Verifiable Finance": A Pragmatic Approach to Decentralization

  • "I think it would be disingenuous to call it decentralized when there's a single permissioned operator who's going to be running the sequencer, which is us... It's pretty clear the market is not a huge fan of max decentralization."
  • Ellipsis Labs rejects the "decentralization at all costs" mindset, opting instead for "verifiable finance." This means while they will run a centralized sequencer for performance, the system’s state transitions are cryptographically guaranteed, and users can always withdraw funds, even if the operator goes rogue.
  • Chen argues that market behavior proves users prioritize performance, liquidity, and user experience over ideological purity. Centralized exchanges still dominate volume for a reason.
  • He notes that many so-called "decentralized" L2s are misleading, as they often have admin keys that give a single entity control over user funds, making them "custodial" in practice.

Fixing Crypto's Incentive Problem

  • "The incentives are set up such that the validator is incentivized to be maximally greedy... On Solana, today the incentive is just to maximally [mess with] the user whenever you can. That is just not conducive to internet capital markets."
  • Solana's current market structure incentivizes validators and other actors to extract value (MEV) by giving traders the worst possible price, a practice known as a "bad fill." This is the opposite of regulated markets, where price improvement is the goal.
  • This dynamic extends to payment for order flow (PFOF). In crypto, wallets or apps can sell their transaction flow to entities that sandwich attack users, whereas in TradFi, PFOF is regulated to ensure users get a better price than on public exchanges.
  • This toxic environment discourages serious capital formation. To achieve the vision of "Internet Capital Markets," crypto must build infrastructure where creating good, efficient fills is the most profitable strategy.

Key Takeaways:

  • Building a better financial system on-chain isn't about simply copying TradFi or chasing maximum decentralization. It's about fundamentally re-architecting market incentives to prioritize performance and fairness, even if it means embracing a more centralized, opinionated design.
  • Stop Chasing Max Decentralization. The market has voted with its volume. Users prioritize performance over ideological purity. "Verifiable Finance"—with centralized sequencers but guaranteed withdrawals—is the pragmatic path forward.
  • Market Structure Is Destiny. Inefficient L1s with toxic MEV force sophisticated teams to build workarounds (like the proprietary AMM Sulfi) or entirely new, controlled environments (like Atlas). The base layer's design dictates the quality of applications built on top.
  • The Real Game Is Efficient Markets, Not Memecoins. The long-term vision for crypto finance depends on building infrastructure that can attract institutional capital with fair, reliable, and highly efficient execution. The current system that incentivizes "bad fills" is a dead end.

For further insights and detailed discussions, watch the full podcast: Link

This episode reveals the critical tension between crypto's decentralization ethos and the pragmatic realities of building a high-performance financial product, offering a masterclass in market structure for Solana investors.

Introduction to Ellipsus Labs and Its Vision

  • Phoenix: A fully on-chain limit order book exchange on Solana that has processed over $75 billion in volume, initially built as a proof-of-concept to demonstrate the superiority of order books for liquid pairs.
  • Gavl: A token launch platform designed for fully on-chain origination, aiming to eliminate the high costs of centralized exchange listings and market makers while minimizing value loss to MEV.
  • Atlas: An opinionated Layer 2 blockchain designed specifically for high-frequency trading, born from the limitations encountered while building Phoenix on Solana's Layer 1.
  • Sulfi: The largest on-chain market maker on Solana, developed by Ellipsus Labs to compete effectively within the current market structure.

The Genesis of Atlas: Why Solana L1 Isn't Enough

Eugene explains that Phoenix was a proof-of-concept to prove that a limit order book—a system that matches buy and sell orders at specified prices—could provide higher quality liquidity than an AMM (Automated Market Maker), which uses mathematical formulas to price assets. However, the growth of Solana led to new challenges, such as transaction landing difficulties and the need for fine-grained control over execution.

The primary motivation for building Atlas was the lack of control over the sequencing layer on Solana L1. Eugene highlights the "proposer monopoly," where a single validator controls transaction ordering for a short period. This environment makes it difficult to run an efficient exchange.

“On the layer one, just like some random guy around the world is running your matching engine for 1.6 seconds. They can [mess] with everything they want to [mess] with... it turns out it's like very difficult to run an efficient exchange in that type of environment.”

  • Strategic Insight: The move to build a dedicated L2 like Atlas signals that for high-performance applications, a general-purpose L1 may not be sufficient. Investors should watch for applications that require specialized environments and may migrate to or build their own app-chains.

Multiple Concurrent Leaders and the L2 Trade-off

The conversation addresses potential Solana L1 improvements like multiple concurrent leaders, a proposed upgrade where multiple validators propose blocks simultaneously to reduce the power of any single proposer. Eugene acknowledges this helps mitigate the proposer monopoly but argues it still doesn't offer the same level of control as a centralized sequencer on an L2.

An L2 with a centralized sequencer, like Atlas, can enable features like truly continuous block building, which is impossible on an L1 constrained by block times. This highlights a core trade-off: sacrificing some decentralization for superior application performance.

Crypto vs. TradFi: The Market Efficiency Gap

Eugene, drawing from his understanding of traditional markets, notes that crypto market efficiency varies wildly. While BTC perpetuals on Binance are extremely efficient, on-chain markets often break down due to a more complex "supply chain" for transactions. This chain involves users, dApps, wallets, RPCs, and validators, creating numerous points of failure or manipulation.

The rapid evolution of the on-chain landscape—from dominant apps to validator software—means the "rules of the game" are constantly changing, inherently making the market less efficient than more stable TradFi systems.

The Nuance Beyond the CLOB vs. AMM Debate

While acknowledging the Central Limit Order Book (CLOB) is generally better for liquid pairs, Eugene pushes back on the idea that it's always superior. He criticizes the current discourse as "lazy" and "sloppy," with many projects simply latching onto the CLOB narrative without understanding the underlying microstructure.

He argues that true performance isn't just about prioritizing cancels over trades. It involves deep, nuanced tinkering with the market microstructure to create the best possible application, a focus he believes is missing from many new projects.

Atlas's North Star: Verifiable Finance, Not Maximal Decentralization

Atlas is being built with a clear, opinionated goal: to be the best possible home for Phoenix. This means prioritizing features that benefit market makers to ensure retail users get the best prices. This includes reliably landing oracle updates and prioritizing market maker orders (cancels and places) over taker orders.

Eugene introduces the term “verifiable finance” to describe Atlas's philosophy. This acknowledges that while the sequencer will be centralized (run by Ellipsus Labs), the system's state transitions are cryptographically guaranteed, and users can always withdraw their funds trustlessly.

  • Actionable Insight: Eugene's pragmatic stance challenges the "decentralization at all costs" narrative. He argues that for many users, verifiability and product performance are more important. Investors should assess whether a project's level of decentralization aligns with its stated goals and target user needs.

Why Settle to Ethereum?

When asked why Atlas would settle to Ethereum, Eugene states the importance of the settlement layer is "a little bit overrated." He argues that a settlement layer's primary job is to provide long-term censorship resistance and guarantees for withdrawals if the L2 operator goes offline or becomes malicious.

Ethereum has optimized for these properties (strong censorship resistance, revert protection), making it a good settlement layer. He notes that most other blockchains have not prioritized these features, making them less suitable for this specific role. He also points out that many L2s today don't offer "true settlement," as admin keys could still compromise user funds, a fact verifiable on sites like L2BEAT.

Tackling MEV and the Gavl Experiment

The discussion shifts to MEV (Maximal Extractable Value), which is the profit a validator or block producer can make by reordering, inserting, or censoring transactions. Eugene explains that Ellipsus Labs first designed a sandwich-resistant AMM to protect users but found that order flow originators (like memecoin launchpads) were not incentivized to adopt it.

This led to the creation of Gavl, a platform for token launches that aims to be less extractive. Gavl focuses on fair initial price discovery (e.g., via a Dutch auction) to prevent sniping and graduates tokens to a sandwich-resistant AMM. The goal is to provide a cheaper, more transparent alternative to the current model of bribing centralized exchanges and market makers.

The User Adoption Dilemma: Why Better Isn't Always Better

A key tension emerges: even if you build a technically superior, more user-friendly product, adoption isn't guaranteed. Eugene notes that users are often insensitive to small price differences, similar to how sports bettors tolerate wide spreads or credit card users focus on points instead of fees.

“The user is insensitive to the fee, but like somewhat sensitive to the incentive. And so then you want to maximize fee and maximize incentive even though it's identical.”

  • Strategic Consideration: This insight is crucial for researchers and investors. A project's success may depend less on pure technical efficiency and more on its ability to navigate user psychology, incentives, and distribution channels. The best tech doesn't always win.

The Story of Sulfi: Playing the Game to Win

Eugene reveals the story behind Sulfi, a proprietary, high-performance on-chain market maker. Sulfi was Ellipsus Labs' response to the existing Solana market structure. Rather than trying to fix the L1, they built a tool to compete as effectively as possible within its constraints.

Sulfi provides a simple swap interface for aggregators like Jupiter but uses a sophisticated, private backend to offer more competitive prices than public venues like Phoenix. This structure is a direct result of Jupiter's focus on securing the best possible price for users, which in turn incentivizes the creation of these proprietary liquidity sources.

Reforming Solana's Market Structure

Looking at the bigger picture, Eugene expresses concern over the current crypto market structure, which often incentivizes bad fills. In crypto, payment for order flow (PFOF)—where a wallet or dApp sells its user transaction stream to a third party—can lead to users getting the worst possible price, unlike in TradFi where it's regulated to ensure price improvement.

He argues that the incentive for validators on Solana is to be "maximally greedy," which is not conducive to building robust "internet capital markets." While improvements like better transaction land rates on Solana help, the fundamental incentive problem remains, especially as more applications begin selling their order flow directly.

Conclusion: A Pragmatic Path to Efficient Markets

This discussion highlights the conflict between building theoretically perfect, decentralized systems and creating products that succeed in the current market. Eugene's optimism lies in the belief that as crypto matures and attracts real financial flows, execution quality and efficient market design will become non-negotiable, moving beyond today's speculative games.

Others You May Like