Prop AMM Banner
/Research

Solana’s Proprietary AMM Revolution

24 min read

Many thanks to 0xIchigo and Ramzy Ali for reviewing earlier versions of this work.

Actionable Insights

  • Proprietary AMMs represent a paradigm shift in how liquidity is provisioned on-chain. By embedding trading strategies directly into the runtime, they eliminate the latency of off-chain execution and offer tighter, more competitive pricing that rivals top centralized exchanges. The model has no direct analogue in traditional finance.
  • Proprietary AMMs have emerged specifically on Solana, fueled by structural factors distinct to the network, including its robust retail base, transaction ordering mechanics, aggregator-concentrated order flow, and a high-throughput, low-cost environment suited for high-frequency price updates.
  • Proprietary AMMs differ from traditional AMMs by providing active liquidity. Pricing strategies are continuously adjusted, independent of trade activity, rather than remaining static until liquidity is consumed. This active approach, which in many cases outperforms passive liquidity models such as constant-product or concentrated-liquidity DEXs, signals a broader structural shift in how on-chain liquidity is provided.
  • Over the past 60 days, daily trading volumes across all proprietary AMMs on Solana have consistently exceeded $1 billion.
  • Over the past three months, SOL/USDC swaps averaged around $1.5 billion in daily volume. Proprietary AMMs have consistently captured more than 60% of this market, peaking at 86% on July 5.
  • A critical enabler of proprietary AMMs is lightweight, low-cost oracle updates, which allow quotes to be refreshed far more cheaply than a standard swap. For instance, HumidiFi has optimized its oracle updates to just 143 CUs, over 1,000 times less than that requested by a typical Jupiter aggregator swap.
  • Lightweight oracle updates with minimal CU requirements need only small tips to secure top positions in the Jito auction queue. This gives market makers a cost-effective form of “cancel priority,” enabling them to refresh quotes ahead of taker transactions and reduce exposure to adverse selection.
  • A proprietary AMM expresses its quotes through a pricing curve, which can be updated through a few parameter adjustments, making it far more computationally efficient than managing individual order book quotes. By concentrating liquidity tightly around the latest oracle price, the AMM achieves extremely high capital efficiency.
  • Solana’s proprietary AMMs rely heavily on trade flow from the Jupiter aggregator. In July, aggregator-driven trades accounted for 99.2% of all volume on GoonFi, 97.3% on Zerobro, 92.5% on Orbic, and 88.4% on SolFi. Over 40% of all Solana DEX swap volume flows through aggregators.
  • Because Jupiter dominates Solana’s aggregator market, new proprietary AMMs only need a single integration to gain immediate, wide distribution to non-toxic retail order flow.

Introduction

Over the past several months, Solana’s trading landscape has undergone a dramatic shift. A roster of unfamiliar new DEXs (SolFi, ZeroFi, Tessera, HumdiFi, etc.) has rapidly seized market share, reshaping the structure of on-chain liquidity. Their rapid dominance has left many observers wondering how these platforms emerged seemingly out of nowhere and why they now account for such a large share of trading activity.

The answer lies in a new market-making primitive pioneered on Solana: proprietary AMMs, a blockchain native model with no direct analogue in traditional finance. Proprietary AMMs represent a paradigm shift in how liquidity is provisioned, moving the core trading logic, normally executed on private servers, directly on-chain within Solana programs. Each proprietary AMM program functions as an independent trading venue, operated by its respective market maker. By embedding strategies into the runtime itself, proprietary AMMs eliminate the latency introduced by off-chain execution and deliver tighter, more competitive pricing.

This innovation (an overused term, but in this case not an overstatement) has already transformed Solana’s spot markets, with proprietary AMMs now dominating trading in major pairs. The dynamic they introduce, active liquidity provision that directly competes with and increasingly outperforms passive liquidity models, signals a broader structural change. Market making on Solana is evolving in favor of sophisticated, on-chain-native participants, and away from passive crowdsourced liquidity pools. The implications for users and liquidity providers alike are profound.

Proprietary AMMs are difficult to research, and as a result, very little beyond trading data has been published about them. They primarily operate without frontends, public IDLs, documentation, marketing, or community channels, and unlike public DEXs, they do not allow retail users to provide liquidity. The space is fiercely competitive, and even the slightest efficiency gain can offer an edge, making secrecy the norm, much like with traditional HFT firms. These proprietary AMMs are operated by crypto-native market makers, many of whom have chosen to remain anonymous.

Proprietary AMMs are a Solana-specific development, enabled by a set of structural factors unique to the network:

  • High Retail Participation – Solana’s strong retail user base makes the network an attractive venue for market makers.
  • Transaction Ordering Mechanics – Compute Units (CUs) play a central role in transaction prioritization, creating opportunities for specialized execution strategies.
  • Aggregator Dominance – A single leading aggregator channels much of the retail order flow, providing an accessible one-stop shop for distribution.
  • Developer Strength – Solana boasts a diverse ecosystem of protocols and skilled developers, fostering innovation and experimentation.
  • Low-cost, High-throughput Environment – the fast and cheap price updates required by the proprietary AMMs make the model best suited to blockchains with high TPS.

The terminology for this emerging class of new market makers remains unsettled; some refer to them as “proprietary AMMs,” while others frame them as “dark pools.”

In traditional finance, dark pools are private trading venues where institutions execute large block trades away from public exchanges, keeping their intentions hidden to avoid moving the market. Proprietary trading, meanwhile, refers to firms deploying their own capital rather than client funds.

“Dark” in the context of proprietary AMMs refers less to hidden order flow and more to the opacity of how these AMMs function. Their swap logic is not publicly disclosed, and without a published IDL, it can be challenging to determine how to interact with the programs. Observers may analyze transactions or attempt to reverse-engineer their logic, but even then, the program can be upgraded at any time to block unwanted integrations. This opacity is very much a feature, not a bug, for the operators, as it’s harder for toxic takers to trade against something they can’t fully see or understand.

A third framing is “oracle-based AMMs,” a label early pioneer Lifinity has long used for itself. Here, the defining feature is reliance on oracle feeds to anchor trades to fair market value. Many view rapid, cheap oracle updates as the key breakthrough of this new model of on-chain trading, effectively enabling a lightweight version of a taker speed bump. We’ll dive deeper into these mechanics later in the article.

The term AMM broadly refers to any automated, algorithmic strategy for providing liquidity and facilitating trades. However, in crypto, it has become closely associated with passive liquidity models such as constant-product or concentrated-liquidity DEXs. Proprietary AMMs differ in a key respect as they represent active liquidity. Their pricing strategies are continuously adjusted, independent of trade activity, rather than remaining static until liquidity is taken.

In this article, we reject using the term dark pool, both because of its negative connotations and its association with an existing concept in traditional finance. Instead, we refer to these on-chain market makers as proprietary AMMs.

Proprietary AMMs Operating on Solana

Operating a proprietary AMM demands both deep on-chain market-making expertise and a detailed understanding of Solana’s market microstructure, creating a barrier to entry for new participants. That said, the competitive landscape is evolving rapidly. Market share has shifted meaningfully over short periods, and with the model proving successful, additional new entrants are expected.

As of today, at least seven proprietary AMMs are active on Solana. Since many of these platforms will be referenced throughout this article, we begin with brief profiles of each.

Lifinity

Lifinity was the first to introduce the proprietary AMM model, launching in January 2022. Branding itself as an “Oracle-Based AMM”.

Lifinity stands apart from other prop AMMs in several ways. Unlike most peers, it maintains a public front-end, though its pools are closed to new deposits. Aggregators account for only about 50% of its trading volume, significantly lower than the share seen on competing proprietary AMMs. The protocol also issued its own token, LFNTY, which grants governance rights and entitles holders to a share of protocol revenue. All liquidity in the DEX is protocol-owned and directly backs the LFNTY token.

Beyond the AMM, Lifinity has expanded into adjacent products, including Sandglass, a Pendle-inspired fixed-income protocol, and Flares, a community NFT initiative. The team remains highly engaged with its user base through frequent X updates and regular community calls on Discord. The Linfinity program can be found at the address 2wT8Y…7aD3c.

SolFi

SolFi is run by Ellipsis Labs, a highly respected Solana development team led by Jarry Xiao and Eugene Chen. The team’s portfolio includes several notable projects, including the on-chain order book Phoenix, the token distribution and liquidity bootstrapping platform Gavel, and the upcoming SVM-based, DeFi-focused blockchain Atlas.

The team has described SolFi as the “natural evolution of Phoenix's orderbook markets”, leveraging their deep expertise in on-chain market microstructure. Ellipsis Labs launched SolFi in early November 2024. Its swift rise in market share is one of the key catalysts that has sparked the current wave of interest in proprietary AMMs. Their program can be found at the vanity address SoLFi…ZxyCe.

Obric

Obric V2 is a proprietary AMM developed by a small team that first launched on Aptos before migrating to Solana. The team refers to Obric as a “proactive AMM,” and the protocol was integrated into the Jupiter aggregator in early October 2024, making it one of the earlier proprietary AMMs on Solana. 

Obric stands out for its multi-chain presence, also operating on Sui and Berachain. This cross-chain activity provides additional visibility into their market-making strategies, which we will examine later in this article. They are one of the few proprietary AMMs to operate an active X account under the quirky handle @poor_obric. Their program can be found at the vanity address obriQ…m113y.

Tessera V

Tessera V is operated by Wintermute, one of the most established market makers in crypto. Founded in 2017 by CEO Evgeny Gaevoy, Wintermute has grown into one of the largest algorithmic trading firms in digital assets. The company provides liquidity across most major exchanges and trading venues, offers a wide suite of OTC trading products, and partners with both Web3 projects and traditional financial institutions as they enter the crypto space.

Tessera V was integrated into the Jupiter router in June 2025. The program can be found at the vanity address TessV…3GLQH.

HumidiFi, GoonFi, and ZeroFi

These three AMMs are operated by professional DeFi market-making firms and Solana-native teams that have so far chosen to keep their involvement under the radar. HumidiFi, in particular, has quickly risen to become one of the largest AMMs by trading volume in recent weeks. HumidiFi and GoonFi launched in June, while ZeroFi began earlier in January. Their program addresses are 9H6tu…3q6Rp, goonER…BnS5j and ZERor…4hbZY, respectively.

Proprietary AMMs Trading Data

This section examines on-chain activity for proprietary AMMs, covering overall trading volumes, market share dynamics, SOL–stablecoin flows, and their dependence on trades routed through the Jupiter aggregator.

Market making is a volume game. Over the past 60 days, total daily trading volumes across all proprietary AMMs on Solana have consistently topped $1 billion, ranging from a peak of $1.93 billion to a low of $406 million. Volume trends closely mirror those of the broader Solana DEX market, with activity typically dipping on weekends.

The proprietary AMM market is fast-moving and fiercely competitive, with new entrants quickly challenging incumbents. For example, HumidiFi, launched only in mid-June, has surged to capture 47.1% of all proprietary AMM volume, overtaking SolFi, which held a dominant 61.8% share just two months earlier.

Proprietary AMM dominance is heavily concentrated in a few trading pairs, most notably SOL-stablecoin swaps, with SOL/USDC by far the most active. Over the past three months, SOL/USDC swaps averaged around $1.5 billion in daily volume. Proprietary AMMs have consistently captured more than 60% of this market, peaking at 86% on July 5. HumidiFi has quickly become the leading venue in this segment, accounting for over 25% of recent volume, while newcomer Tessera, which launched in July, has already grown to around 6%.

Jupiter Aggregator 

Since its launch in October 2021, Jupiter has effectively defined the aggregator category on Solana, driven by deep integrations across the ecosystem, a strong social media presence, and unmatched user mindshare. Its flagship swap aggregator anchors a rapidly expanding suite of products designed to position Jupiter as an all-in-one crypto super-app.

The team has rolled out native offerings such as Jupiter Perps, JLP, the Jupiter mobile app, DCA, Ape Pro, the Jupiter Studio launchpad, and Jupiter Lend (built in collaboration with Fluid). At the same time, they’ve pursued an aggressive acquisition strategy, bringing in projects such as Moonshot, Drip Haus, Solana FM, and Sonar Watch.

Over 40% of all Solana DEX swap volume flows through aggregators, with Jupiter firmly dominating the space. In July 2025 alone, Jupiter captured 86.4% of aggregator activity, $55.4 billion out of $64.1 billion in total swap volume.

This market structure stands in contrast to that of other mature networks. On Ethereum, for instance, the DEX aggregator landscape is more diversified, with no single player holding Jupiter-like dominance. In July, Ethereum aggregators processed $31.09 billion in trades, led by Cow Swap with $9.07 billion (29.2%), followed by 1inch at $7.72 billion (24.8%) and Kyber at $3.77 billion (12.1%). These market share proportions have remained roughly stable over the past year, as shown in the chart below.

Solana’s proprietary AMMs rely heavily on trade flow from Jupiter, far more so than other DEXs, which tend to have a broader mix of transaction sources. In July, aggregator-driven trades accounted for 99.2% of all volume on GoonFi, 97.3% on ZeroFi, 92.5% on Orbic, and 88.4% on SolFi. By comparison, leading Solana DEXs such as Raydium, Meteora, and Orca saw far lower reliance: 19.4%, 21.4%, and 20.3% respectively.

However, this wasn’t always the case. Historical data shows that, prior to SolFi’s launch in late 2024, a much larger share of Raydium, Meteora, and Orca volume was routed through aggregators. At that time, 63.7% of Meteora’s volume, 49.3% of Orca’s, and 32.6% of Raydium’s came via Jupiter. This decline suggests that Prop AMMs have been steadily winning a greater share of Jupiter-sourced trades from their public DEX counterparts.

Jupiter aggregator volume data confirms this trend. Since its launch late last year, SolFi alone has quickly grown to capture a significant share of all trades routed through Jupiter. In early May, its volume peaked at 46.4% of Jupiter’s total USD-denominated flow, nearly half the market. Although new entrants have eroded this share, SolFi remains one of the largest venues in the proprietary AMM space. By early August, SolFi and ZeroFi accounted for 18.4% and 13.9% of volume, respectively.

How do proprietary AMMs work?

In this section, we take a closer look at the inner workings of proprietary AMMs, focusing on key components such as oracle updates, inventory management, and their interaction with the Jupiter aggregator.

Before understanding the advantages that proprietary AMMs bring to professional on-chain market makers, it’s important to examine the traditional alternative: order books. On Solana, these are represented by protocols such as Phoenix and OpenBook v2. While these venues enable market makers to provide spot liquidity in a familiar limit order book format, the on-chain model imposes significant operational and economic challenges.

Maintaining active quotes on Solana order books is computationally expensive. Updating orders on Phoenix, for instance, consumes 100,000–300,000 compute units (CUs) per transaction, with similar ranges seen across other DEXs. Because every quote, update, or cancellation requires a complex transaction, market makers face high costs simply to keep their liquidity live. Makers must pay tips or priority fees to ensure their updates land quickly enough to remain competitive, which eats into profits.

On-chain order books broadcast every quote to the network. This transparency creates exploitable opportunities for latency-sensitive actors who can snipe stale orders during volatile price moves. For market makers, this means a greater risk of adverse selection and higher costs to defend quotes, especially when spreads are tight.

It is no accident that SolFi, developed by Ellipsis Labs, was the team to refine Lifinity’s pioneering model and drive the broader adoption of proprietary AMMs. As both the creators of Phoenix and active market makers on it, Ellipsis Labs developed firsthand insight into the limitations of the order book model within Solana’s technical constraints.

A market maker’s quoting algorithm is built around two key components:

  • Fair Value Estimation – a way to determine the current fair price of the token pair.
  • Execution Logic (making strategy) – translating that fair value into actionable quotes. This logic incorporates factors such as inventory balance and risk management preferences.

In both on-chain order books and proprietary AMM models, a market maker’s fair value estimation for tokens is sourced off-chain, since price discovery occurs on the most liquid venues such as Binance. The key distinction lies in execution, where the proprietary AMM model enables market makers to bring their execution logic on-chain by deploying a custom independent program under their control, effectively embedding their market-making strategy directly into the blockchain.

To achieve this efficiently, a market maker’s willingness to trade across prices is represented not through a series of discrete limit orders, but as a continuous mathematical curve. Updating this curve requires only a small set of parameter adjustments, making it far more computationally efficient than managing individual quotes on an order book.

This use of curves gives proprietary AMMs a closer resemblance to more traditional concentrated liquidity AMMs. Given that proprietary AMMs are essentially black boxes, the precise functions employed are not observable, though some have developed simulators to observe their behavior.

Oracle Updates

The key innovation enabling proprietary AMMs to place their logic on-chain is lightweight and low-cost oracle updates. Proprietary AMMs leverage lightweight developer frameworks like Pinocchio or even sBPF Assembly to minimize the compute required for updating parameters, allowing them to refresh quotes at a fraction of the cost of a standard swap.

An example is HumidiFi, which has optimized its oracle updates to consume just 143 CUs, which means an update transaction costs just 9,998 lamports ($0.001784). In comparison, a typical Jupiter aggregator swap routed through a proprietary AMM typically consumes around 150,000 CUs, over 1,000 times more compute.

This disparity matters because the Jito auction engine, through which time-sensitive transactions are run, prioritizes transactions by tip per CU. Lightweight oracle updates with tiny CU requirements only need modest tips (in HumidFi’s case, 4,998 lamports) to rise to the top of the auction queue, giving market makers a powerful form of “cancel priority.” Allowing them to update their quotes ahead of taker transactions and reducing exposure to adverse selection. 

This ability to update cheaply and frequently enables market makers to quote extremely tight spreads on-chain with less fear of being picked off by toxic flow. Since these systems maintain minimal state, market makers can update multiple markets simultaneously. It’s also possible to add additional safeguards: for instance, rejecting any taker transaction that consumes fewer than 100,000 CUs, on the basis that a cheaper swap attempt is more likely to be an exploitative one. Oracle updates are sent on-chain by a designated authority account that pushes curve adjustments multiple times per second. If this authority stops updating for whatever reason, swaps in that pool likely halt.

As competition intensifies, the dominant proprietary AMMs are in a race toward minimal compute usage and maximal update frequency.

Execution Strategies

As discussed previously, the proprietary AMM Obric is also deployed on networks including Sui. The Move programming language used on Sui retains more information in its bytecode than many alternatives, making program decompilation and analysis significantly easier. This transparency gives us valuable insight into the inner workings of Obric’s AMM execution logic.

An analysis conducted by Marko of Mysten Labs of Obric’s Sui program reveals that the AMM strategically provides tightly concentrated liquidity around oracle prices. Instead of relying on a fixed-product formula like in traditional constant-product AMMs (i.e., xy = k), Obric adjusts its effective k dynamically based on several parameters:

  • target_x: The desired quantity of token X, determined by the amount deposited in the pool.
  • concentration: A configurable parameter that dictates how tightly liquidity is clustered around the price point.
  • mult_x / mult_y: Oracle-derived price multipliers that further shape the liquidity curve

By setting target amounts for each token in the pool and tuning concentration around the oracle price, Obric maintains an inventory that remains close to its optimal balance, while still offering high capital efficiency through targeted liquidity.

Below is the TradingPair struct from the decompiled program.

Code
struct TradingPair<phantom TokenX, phantom TokenY> has key {
    id: 0x2::object::UID,
    reserve_x: 0x2::balance::Balance<TokenX>,
    reserve_y: 0x2::balance::Balance<TokenY>,
    concentration: u64,        // Liquidity concentration parameter
    big_k: u128,               // Constant product invariant
    target_x: u64,             // Target reserve for token X
    mult_x: u64,               // Price multiplier for token X
    mult_y: u64,               // Price multiplier for token Y
    fee_millionth: u64,        // Fee in parts per million
    x_price_id: address,       // Price feed ID for token X
    y_price_id: address,       // Price feed ID for token Y
    x_retain_decimals: u64,    // Decimal precision for token X
    y_retain_decimals: u64,    // Decimal precision for token Y
    cumulative_volume: u64,    // Tracking volume
    volumes: vector<u64>,      //
    times: vector<u64>,        // Trade timestamps?
    target_y_based_lock: bool, // Circuit breaker?
    target_y_reference: u64,   //
    pyth_mode: bool,           // Whether to use Pyth oracle prices
    pyth_y_add: u64,           // Manual offset on pyth prices?
    pyth_y_sub: u64,
}

It is important to note that execution logic is not standardized, and other proprietary AMMs likely take different approaches than Obric. What is consistent, however, is the use of concentrated liquidity tightly clustered around the latest oracle price, enabling proprietary AMMs to achieve very high capital efficiency.

As noted by Benedict Brady, founder of Meridian and former quantitative researcher at DeFi-focused trading firm Ergonia, proprietary AMMs give operators far greater flexibility in how they design liquidity. They can experiment with alternative pricing curves, embed custom logic into swap mechanics, and effectively define their own liquidity models, offering significantly more degrees of freedom than traditional approaches.

Inventories

Proprietary AMMs operate without external liquidity providers, instead managing their own inventory through program-controlled vaults, which are called upon to fund trades. These vaults effectively act as the AMM’s balance sheet.

Proprietary AMMs take on the price risk of the assets they hold in inventory, making this model less suited to long-tail, high-volatility assets such as new memecoins. While this exposure can be hedged through short positions elsewhere, doing so lowers capital efficiency and adds operational complexity, as vault balances fluctuate with trading activity. Furthermore, price discovery for long-tail assets occurs primarily on-chain, diminishing the relevance of an oracle-based model.

Accordingly, many proprietary AMMs limit their quoting to frequently traded liquid trading pairs where they can close the spread. Having said this, as competition intensifies and the proprietary AMM model evolves, operators are broadening the set of tokens they quote, moving beyond just SOL-stablecoin and stablecoin–stablecoin pairs. For example, ZeroFi maintains trading inventory across a range of assets, including majors like SOL, USDC, CbBTC, and wETH, Solana blue-chip DeFi governance tokens such as JUP and Kamino, as well as established memecoins like Popcat, WIF, and Bonk. Proprietary AMMs are also scaling to handle larger trade sizes, as demonstrated on August 13 when SolFi and Tessera executed an on-chain swap of $4M USDC into SOL.

Jupiter Router

As detailed in the data section of this article, proprietary AMMs source the majority of their flow through the Jupiter aggregator. This flow is highly attractive as it is predominantly clean, non-toxic retail order flow, which makes it easier for market makers to quote tightly with confidence. Because Jupiter dominates Solana’s routing layer, new proprietary AMMs only need a single integration to gain immediate distribution.

This market structure exists because Jupiter is intensely focused on execution quality. Even though retail traders are not sensitive to single-basis-point differences, Jupiter consistently routes trades to deliver the best available price. To achieve this, the platform combines multiple routers. The latest version of its proprietary DEX aggregation engine, Metis v1.6, is the router used by proprietary AMMs. Jupiter also integrates other sources, including its RFQ system Jupiter Z, and acts as a meta-aggregator for services such as Hashflow, DFlow, Pyth Express Relay, and the OKX DEX Router.

As of writing, the Jupiter Metis router supports 56 on-chain DEXs. Integration is permissioned but does not carry fees. When assessing whether to onboard a new DEX, the team considers factors like code quality, security audits, product traction, and the caliber of the team and its backers. AMMs must provide Jupiter with an SDK compatible with the Jupiter AMM Interface. According to Jupiter co-founder Siong, he has personally met with every proprietary AMM operator and ensures that a trusted auditor reviews both sides of any integration.

From the end user’s perspective, unless they check their transaction through a block explorer, proprietary AMMs are invisible. Traders submit their orders through a front-end, and the aggregator automatically selects the optimal route, which may or may not include one or more prop AMMs. Users are indifferent to the venue; what matters is execution quality—tight quotes and reliable fills. In this structure, aggregators effectively assume the role of brokers, holding significant influence over the downstream market makers.

When swapping through Jupiter, retail users can choose between two modes: Ultra V2 (the default) or Manual. Ultra V2 includes more advanced features such as real-time slippage estimation, which automatically adjusts slippage tolerance to optimize trade execution, a key benefit for less experienced traders. The Manual option gives users full control over swap settings, including priority fees, slippage, and broadcast mode. Jupiter charges a 10 or 5 basis point fee on Ultra V2 swaps, depending on the tokens involved (select stable swaps are free), while Manual swaps incur no fees. This structure is highly competitive compared to other major retail flow providers. Phantom, for instance, charges 85 basis points on wallet-based swaps.

A key nuance is that the Jupiter router also supports a swap API, which opens the door for professional traders to mimic retail flow in an attempt to pick off stale quotes. As a result, while aggregators filter much of the toxic order flow, proprietary AMMs must still exercise caution.

New Market Structure

By embedding their core trading logic directly into on-chain programs, proprietary AMMs have redefined what it means to execute algorithmic strategies on a blockchain. Unlike traditional models where strategies are hosted off-chain and only settlement occurs on-chain, these programs operate natively within Solana’s runtime, executing pre-programmed strategies in real time with ultra-low latency.

This shift was not always feasible. The prevailing trend across blockchains in earlier years was to move execution logic off-chain. Applications relied on off-chain matching engines, with only settlement recorded on-chain, because blockchains lacked the compute and throughput to support more complex on-chain strategies. Solana’s performance changes this dynamic, and it becomes possible to keep strategy execution within the chain itself, restoring the model to what some argue is its “correct” form.

As a result, Solana’s AMM ecosystem has bifurcated into two distinct classes:

  • Generalized Public DEXs (passive liquidity) such as Orca and Raydium, which cater to the long tail of assets, retail traders, and community liquidity providers.
  • Proprietary AMMs (active liquidity), private and often opaque venues optimized for highly liquid, well-established assets, where professional market makers deploy their own capital and execution logic.

While this model unlocks efficiency for sophisticated participants and better pricing for retail users, it also could raise concerns if liquidity and order flow become highly concentrated within a small set of closed-source entities, limiting transparency and potentially reinforcing centralization.

This structure also cements Jupiter’s central role as Solana’s dominant liquidity aggregator. By routing the majority of order flow, Jupiter effectively acts as the arbiter of access: those with integration into its system gain privileged visibility and liquidity, while those without risk marginalization. The aggregator’s influence gives it the ability to set the rules of the road for downstream market makers, underscoring the power dynamics of Solana’s current market structure.

It should be noted that there are no indications of Jupiter unfairly gatekeeping access to order flow. Looking ahead, it is expected that proprietary AMMs will naturally seek to further diversify their integration with additional sources of order flow.

Conclusion

The emergence of proprietary AMMs marks a fundamental shift in on-chain market making, one that is native to blockchains and without a direct analogue in traditional finance. Rather than replicating legacy market structures, these systems are charting their own path.

This development highlights how Solana’s low-cost, high-throughput design uniquely enables new forms of financial innovation. What stands out is that this shift isn’t the result of top-down direction or formal academic research, but is instead arising organically from competitive market forces and a decentralized community of developers eager to innovate.

Further Resources

Related Articles

Subscribe to Helius

Stay up-to-date with the latest in Solana development and receive updates when we post