Subscribe to Gelora Research
<100 subscribers
In less than two years Hyperliquid, went from a nascent idea to capturing a staggering 30% of Binance's open interest and achieving a valuation comparable to nearly half of Solana's entire fully diluted value. Its success did not stem from a radical new invention, but from the relentless execution of a simple, powerful idea: delivering a CEX-like trading experience on a decentralized, self-custodial platform. By building its own purpose-built L1 blockchain, Hyperliquid achieved the low latency and high throughput necessary to support a real-time order book—a Holy Grail that had long eluded DEXs on general-purpose chains
When trader James Wynn reportedly lost over $100 million on the platform, it became a paradoxical advertisement. The industry saw proof of what it craved: a venue with real, substantial "highly profitable taker flow" —retail and speculative traders willing to cross the spread. This turned Hyperliquid into a honeypot that institutional market makers could no longer ignore
In less than two years Hyperliquid, went from a nascent idea to capturing a staggering 30% of Binance's open interest and achieving a valuation comparable to nearly half of Solana's entire fully diluted value. Its success did not stem from a radical new invention, but from the relentless execution of a simple, powerful idea: delivering a CEX-like trading experience on a decentralized, self-custodial platform. By building its own purpose-built L1 blockchain, Hyperliquid achieved the low latency and high throughput necessary to support a real-time order book—a Holy Grail that had long eluded DEXs on general-purpose chains
When trader James Wynn reportedly lost over $100 million on the platform, it became a paradoxical advertisement. The industry saw proof of what it craved: a venue with real, substantial "highly profitable taker flow" —retail and speculative traders willing to cross the spread. This turned Hyperliquid into a honeypot that institutional market makers could no longer ignore
This explosive growth served as a powerful indictment of the previous generation of DEXs, particularly AMMs where liquidity providers consistently "bleed money" through impermanent loss and sandwich attacks.Hyperliquid's success effectively fired the starting gun for what has been dubbed the CLOB Wars. But this is not merely a competition between applications. On this proxy war for a fundamental architectural future, leading the charge for a revolutionary new approach has emerged:
CLOBs on Blobs — this model combines the off-chain performance of a time-tested CLOB with the scalable infrastructure of modular DA layers, or blobs, all made verifiable and trustless through the cryptographic power of ZK proofs
The emergence of this paradigm split the industry into several camps.
On one side stand the monolithic L1 champions, championed by figures like Solana co-founder Anatoly Yakovenko (@aeyakovenko). Their philosophy is to create a single, hyper-optimized L1 so fast that any CLOB can run on it natively. Their core argument concerns value accrual. As Yakovenko pointedly asked, "But how would solana stakers make any rev from a clob with a network extension?". He argues that rollups, by processing transactions off-chain and only paying low fees for data posting, siphon away valuable economic activity that would otherwise be captured by L1 validators.
On the other side is the modular vanguard, the builders of the CLOBs on Blobs themselves. They argue that the monolithic approach is "fundamentally flawed for high-performance finance". Their "value accrual fallacy" counter-argument is that rollups do not leak value; they create entirely new economic activity that could never exist on the L1 in the first place. The intense transactional load of a vibrant CLOB, with millions of orders per day, would simply overwhelm a general-purpose chain. The real value leakage, they contend, is the user base and capital that is "literally bridging to Hyperliquid instead because the current alternatives on Solana don't cut it". From this perspective, the data availability fees paid by a successful rollup are net-new revenue for the L1, capturing a market it was previously unable to service.
This camp includes a formidable roster of other technically sophisticated teams, such as Bullet (a ZK-rollup/Network Extension on Solana) , the privacy-focused Hibachi(powered by Celestia and Succinct), Lighter (a ZK-rollup on Ethereum) , and Paradex (a Starknet appchain).
The CLOB Wars are therefore a microcosm of the industry's most significant architectural schism: Monolithic vs. Modular. The success or failure of the next generation of DEXs, built on the CLOBs on Blobs model, will provide a powerful, market-validated data point in this ongoing debate, shaping the flow of capital and developer talent for years to come.
CLOBs on Blobs model represents a radical departure from the past. It’s an act of architectural unbundling, deconstructing the monolithic world computer into a set of specialized, interoperable layers that are each hyper-optimized for a single function: execution, settlement, and data availability. To understand how this new machine achieves CEX-level performance without sacrificing the crypto ethos of trustlessness, we must dissect its anatomy piece by piece.
CEX-Performance via Off-Chain Execution
The journey of a trade begins not on a public, congested L1 mempool, but in a private, high-performance environment. An order is sent directly to a dedicated, off-chain sequencer that runs a matching engine on a standard server, much like a centralized exchange. This is the system's engine room, where an order to buy or sell can be received, matched, and executed in microseconds or single-digit milliseconds. This off-chain execution is the source of the CEX-like latency that is essential for attracting professional traders and market makers who need to react to market movements instantly.
This raises the immediate and crucial question: if the execution is happening off-chain in a centralized environment, how can it be trusted? The answer lies in the next two layers.
Verifiable Data on Blobs
To remain trust-minimized, the sequencer must publicly commit to every action it takes. It does this by taking thousands of individual state changes—matched trades, new limit orders, cancellations—and bundling them into a single, compressed batch. This entire batch of raw data is then published to a modular DA layer.
This is where the Blobs come in. The term, popularized by Ethereum's EIP-4844 upgrade , represents a new type of cheap, abundant data space offered by specialized DA layers like Celestia or EigenDA. The DA layer's only job is to guarantee that this data is available for anyone to inspect; it does not execute or interpret it. This allows any independent third party to download the data and verify that the sequencer has not engaged in fraud, such as stealing funds or unfairly censoring users. This symbiotic relationship between the off-chain engine and the on-chain data ledger is the foundation of the model's security.
ZK-Proofs as the Linchpin of Trust
Publishing data ensures transparency, but it doesn't guarantee correctness. Forcing every node in a decentralized network to re-execute thousands of transactions from the blob data would be slow and defeat the purpose of the off-chain engine. The true innovation of the modern CLOB architecture is the use of ZK proofs to bridge this gap.
Think of the ZK proof as a cryptographic notary. After processing a batch, the sequencer uses a zkVM to generate a small ZK proof. This ZKP is a mathematical guarantee that certifies that the thousands of transactions in the batch were executed correctly according to the protocol's rules, all without revealing any of the private, intermediate data.
The emergence of developer-friendly zkVMs like Succinct's SP1 is a critical enabler for this architecture. Historically, creating ZK proofs required deep cryptographic expertise. Now, developers can write their complex CLOB matching engine in a standard language like Rust, and the zkVM handles the heavy lifting of generating a proof of its correct execution.
"A zkVM makes encoding the arbitrary logic as easy as writing a normal Rust program"
On-Chain Verification and Settlement
The final step is to anchor this entire process to a secure settlement layer, such as Ethereum or Solana. The sequencer submits the small ZK proof to a verifier smart contract deployed on the L1. This contract performs a single, computationally cheap operation: it verifies the proof's validity.
If the proof is valid, the contract is cryptographically assured that the entire off-chain batch was executed correctly. It then updates the on-chain state of the rollup, for example, by updating the token balances in the main bridge contract to reflect the net result of all trades. This is the moment of hard finality, where transactions become irreversible and inherit the full security of the base layer.
While the CLOBs on Blobs architecture offers a compelling path forward, its implementation is fraught with significant challenges and trade-offs. The pursuit of CEX-level performance introduces new centralization vectors and complex security models that every project must navigate. Understanding these obstacles is key to evaluating the viability and long-term defensibility of the contenders in the CLOB Wars.
Centralized sequencer
The most significant architectural compromise made by high-performance rollups is the use of a single, centralized sequencer. This entity, responsible for ordering transactions, executing them, and posting proofs, is the source of the single-digit millisecond latency that traders demand. However, this centralization is also the rollup's greatest vulnerability.
It introduces a critical risk: transaction censorship. This occurs when a privileged actor, like a sequencer, deliberately refuses to include a user's valid transaction. In the context of a CLOB, a malicious sequencer could inflict significant financial harm by ignoring a user's attempt to cancel a stale order or, most critically, withdraw their funds from the exchange. This is not a theoretical risk; it is the primary reason decentralization matters.
The Escape Hatch
Given the risk of a centralized sequencer becoming malicious or simply failing, a robust security mechanism is required. This mechanism is the escape hatch, a trust-minimized protocol that allows a user to withdraw their assets directly from the rollup's smart contracts on the settlement layer (L1), completely bypassing the L2 operator. It is the ultimate guarantee of self-custody. There are two primary forms:
Forced Inclusion: This is the main defense against an online but censoring sequencer. It allows a user to submit their transaction directly to a contract on the L1, which then forces the sequencer to include it after a time delay.
Forced Withdrawal: This is the primary defense against a completely failed or offline operator. A user can present a cryptographic proof of their assets on the L2 directly to the L1 bridge contract and withdraw their funds. For this to work, the rollup's transaction data must be available on a DA layer, reinforcing the critical importance of the "Blobs" in the architecture.
The robustness of this escape hatch is a crucial differentiator. It is a fundamental security proposition for rollup-based DEXs like Bullet, Hibachi, and GTE. Conversely, L1-native CLOBs like Hyperliquid and Kuru do not have this concept; their security relies entirely on the decentralization of their own validator sets, which represents a starkly different trust assumption.
Two Finalities
In the world of rollups, the concept of "finality"—the moment a transaction becomes irreversible—exists on a spectrum. Understanding this distinction is essential for managing risk.
Soft Finality: This is the near-instant confirmation a user receives directly from the rollup's sequencer, often in mere milliseconds. It is a "soft" promise that the transaction will be included in the next batch. This provides the responsive, CEX-like user experience necessary for active trading, but it relies on trusting the sequencer is operating honestly.
Hard Finality: This is the point at which a transaction becomes truly irreversible, secured by the full consensus of the underlying settlement layer. This occurs only after the batch containing the transaction and its ZK proof has been posted and verified on the L1. This process is far more secure but also significantly slower, taking from a few seconds for rollups on Solana to ~13 minutes on Ethereum or even hours for complex L3s like Paradex.
This duality has direct consequences. Traders can and should rely on soft finality for low-risk, time-sensitive actions like placing or canceling orders. However, for high-value, security-critical actions like depositing or withdrawing large amounts of capital, users must wait for hard finality.
Composability
A primary and valid criticism of the Layer 2 rollup model is the sacrifice of atomic composability—the ability to interact with multiple independent protocols within a single, all-or-nothing transaction. When a CLOB moves to its own rollup, it becomes an "execution silo," unable to atomically compose with protocols on the L1. This fragmentation is a significant drawback for a generalized DeFi ecosystem built on "money legos".
However, proponents of specialized rollups make a compelling counterargument, particularly for perpetual futures markets. The dominant thesis, validated by the immense success of isolated platforms like Binance and Hyperliquid, is that perpetuals exchanges do not require deep, synchronous composability to thrive. They can and do function as highly successful "vertically integrated platforms". The calculus may be different for spot markets, which, as the Kuru team argues, are more foundational to a general-purpose DeFi ecosystem and "cannot" live in isolation.
The modular, ZK-powered rollup approach—combining high-performance off-chain execution, verifiable zero-knowledge proofs, and scalable data availability layers—solves fundamental issues that have historically plagued on-chain trading.
Hyperliquid's explosive success validated this model, simultaneously highlighting the inherent limitations of monolithic architectures like Solana. Meanwhile, specialized rollups like Bullet, Hibachi, and Lighter have demonstrated superior security guarantees, robust escape hatch mechanisms, and dramatically improved performance metrics.
Yet significant challenges remain: bootstrapping deep liquidity, simplifying the user experience, and creating transparent, fair forced-exit mechanisms are critical hurdles that must be overcome. These are not mere technical details but foundational requirements for mass adoption.
Looking forward, several trends will shape the future of the space:
Decentralizing Sequencers: Achieving performant yet genuinely decentralized sequencing will be essential to securing trust and long-term viability.
Cross-Chain Liquidity: Platforms that seamlessly aggregate liquidity across multiple chains will dominate the increasingly fragmented multi-chain landscape.
Privacy and Sovereignty: Privacy-enhanced platforms utilizing ZK technology will capture significant institutional and sophisticated retail market segments.
Blurring of L1/L2 Boundaries: The distinction between layers will continue to fade, with the most successful platforms defined by their combined performance, security, and user experience rather than their technical labels.
Ultimately, the CLOB Wars will not be won solely by superior technology. Victory will belong to platforms that synthesize the best aspects of centralized exchanges—deep liquidity, intuitive user experience, robust risk management—with the decentralization, transparency, and composability that only blockchain-based systems can offer. The race is fierce, innovation is rapid, and the ultimate winner is clear: the users, who stand to benefit from an open, performant, and truly decentralized financial system.
This explosive growth served as a powerful indictment of the previous generation of DEXs, particularly AMMs where liquidity providers consistently "bleed money" through impermanent loss and sandwich attacks.Hyperliquid's success effectively fired the starting gun for what has been dubbed the CLOB Wars. But this is not merely a competition between applications. On this proxy war for a fundamental architectural future, leading the charge for a revolutionary new approach has emerged:
CLOBs on Blobs — this model combines the off-chain performance of a time-tested CLOB with the scalable infrastructure of modular DA layers, or blobs, all made verifiable and trustless through the cryptographic power of ZK proofs
The emergence of this paradigm split the industry into several camps.
On one side stand the monolithic L1 champions, championed by figures like Solana co-founder Anatoly Yakovenko (@aeyakovenko). Their philosophy is to create a single, hyper-optimized L1 so fast that any CLOB can run on it natively. Their core argument concerns value accrual. As Yakovenko pointedly asked, "But how would solana stakers make any rev from a clob with a network extension?". He argues that rollups, by processing transactions off-chain and only paying low fees for data posting, siphon away valuable economic activity that would otherwise be captured by L1 validators.
On the other side is the modular vanguard, the builders of the CLOBs on Blobs themselves. They argue that the monolithic approach is "fundamentally flawed for high-performance finance". Their "value accrual fallacy" counter-argument is that rollups do not leak value; they create entirely new economic activity that could never exist on the L1 in the first place. The intense transactional load of a vibrant CLOB, with millions of orders per day, would simply overwhelm a general-purpose chain. The real value leakage, they contend, is the user base and capital that is "literally bridging to Hyperliquid instead because the current alternatives on Solana don't cut it". From this perspective, the data availability fees paid by a successful rollup are net-new revenue for the L1, capturing a market it was previously unable to service.
This camp includes a formidable roster of other technically sophisticated teams, such as Bullet (a ZK-rollup/Network Extension on Solana) , the privacy-focused Hibachi(powered by Celestia and Succinct), Lighter (a ZK-rollup on Ethereum) , and Paradex (a Starknet appchain).
The CLOB Wars are therefore a microcosm of the industry's most significant architectural schism: Monolithic vs. Modular. The success or failure of the next generation of DEXs, built on the CLOBs on Blobs model, will provide a powerful, market-validated data point in this ongoing debate, shaping the flow of capital and developer talent for years to come.
CLOBs on Blobs model represents a radical departure from the past. It’s an act of architectural unbundling, deconstructing the monolithic world computer into a set of specialized, interoperable layers that are each hyper-optimized for a single function: execution, settlement, and data availability. To understand how this new machine achieves CEX-level performance without sacrificing the crypto ethos of trustlessness, we must dissect its anatomy piece by piece.
CEX-Performance via Off-Chain Execution
The journey of a trade begins not on a public, congested L1 mempool, but in a private, high-performance environment. An order is sent directly to a dedicated, off-chain sequencer that runs a matching engine on a standard server, much like a centralized exchange. This is the system's engine room, where an order to buy or sell can be received, matched, and executed in microseconds or single-digit milliseconds. This off-chain execution is the source of the CEX-like latency that is essential for attracting professional traders and market makers who need to react to market movements instantly.
This raises the immediate and crucial question: if the execution is happening off-chain in a centralized environment, how can it be trusted? The answer lies in the next two layers.
Verifiable Data on Blobs
To remain trust-minimized, the sequencer must publicly commit to every action it takes. It does this by taking thousands of individual state changes—matched trades, new limit orders, cancellations—and bundling them into a single, compressed batch. This entire batch of raw data is then published to a modular DA layer.
This is where the Blobs come in. The term, popularized by Ethereum's EIP-4844 upgrade , represents a new type of cheap, abundant data space offered by specialized DA layers like Celestia or EigenDA. The DA layer's only job is to guarantee that this data is available for anyone to inspect; it does not execute or interpret it. This allows any independent third party to download the data and verify that the sequencer has not engaged in fraud, such as stealing funds or unfairly censoring users. This symbiotic relationship between the off-chain engine and the on-chain data ledger is the foundation of the model's security.
ZK-Proofs as the Linchpin of Trust
Publishing data ensures transparency, but it doesn't guarantee correctness. Forcing every node in a decentralized network to re-execute thousands of transactions from the blob data would be slow and defeat the purpose of the off-chain engine. The true innovation of the modern CLOB architecture is the use of ZK proofs to bridge this gap.
Think of the ZK proof as a cryptographic notary. After processing a batch, the sequencer uses a zkVM to generate a small ZK proof. This ZKP is a mathematical guarantee that certifies that the thousands of transactions in the batch were executed correctly according to the protocol's rules, all without revealing any of the private, intermediate data.
The emergence of developer-friendly zkVMs like Succinct's SP1 is a critical enabler for this architecture. Historically, creating ZK proofs required deep cryptographic expertise. Now, developers can write their complex CLOB matching engine in a standard language like Rust, and the zkVM handles the heavy lifting of generating a proof of its correct execution.
"A zkVM makes encoding the arbitrary logic as easy as writing a normal Rust program"
On-Chain Verification and Settlement
The final step is to anchor this entire process to a secure settlement layer, such as Ethereum or Solana. The sequencer submits the small ZK proof to a verifier smart contract deployed on the L1. This contract performs a single, computationally cheap operation: it verifies the proof's validity.
If the proof is valid, the contract is cryptographically assured that the entire off-chain batch was executed correctly. It then updates the on-chain state of the rollup, for example, by updating the token balances in the main bridge contract to reflect the net result of all trades. This is the moment of hard finality, where transactions become irreversible and inherit the full security of the base layer.
While the CLOBs on Blobs architecture offers a compelling path forward, its implementation is fraught with significant challenges and trade-offs. The pursuit of CEX-level performance introduces new centralization vectors and complex security models that every project must navigate. Understanding these obstacles is key to evaluating the viability and long-term defensibility of the contenders in the CLOB Wars.
Centralized sequencer
The most significant architectural compromise made by high-performance rollups is the use of a single, centralized sequencer. This entity, responsible for ordering transactions, executing them, and posting proofs, is the source of the single-digit millisecond latency that traders demand. However, this centralization is also the rollup's greatest vulnerability.
It introduces a critical risk: transaction censorship. This occurs when a privileged actor, like a sequencer, deliberately refuses to include a user's valid transaction. In the context of a CLOB, a malicious sequencer could inflict significant financial harm by ignoring a user's attempt to cancel a stale order or, most critically, withdraw their funds from the exchange. This is not a theoretical risk; it is the primary reason decentralization matters.
The Escape Hatch
Given the risk of a centralized sequencer becoming malicious or simply failing, a robust security mechanism is required. This mechanism is the escape hatch, a trust-minimized protocol that allows a user to withdraw their assets directly from the rollup's smart contracts on the settlement layer (L1), completely bypassing the L2 operator. It is the ultimate guarantee of self-custody. There are two primary forms:
Forced Inclusion: This is the main defense against an online but censoring sequencer. It allows a user to submit their transaction directly to a contract on the L1, which then forces the sequencer to include it after a time delay.
Forced Withdrawal: This is the primary defense against a completely failed or offline operator. A user can present a cryptographic proof of their assets on the L2 directly to the L1 bridge contract and withdraw their funds. For this to work, the rollup's transaction data must be available on a DA layer, reinforcing the critical importance of the "Blobs" in the architecture.
The robustness of this escape hatch is a crucial differentiator. It is a fundamental security proposition for rollup-based DEXs like Bullet, Hibachi, and GTE. Conversely, L1-native CLOBs like Hyperliquid and Kuru do not have this concept; their security relies entirely on the decentralization of their own validator sets, which represents a starkly different trust assumption.
Two Finalities
In the world of rollups, the concept of "finality"—the moment a transaction becomes irreversible—exists on a spectrum. Understanding this distinction is essential for managing risk.
Soft Finality: This is the near-instant confirmation a user receives directly from the rollup's sequencer, often in mere milliseconds. It is a "soft" promise that the transaction will be included in the next batch. This provides the responsive, CEX-like user experience necessary for active trading, but it relies on trusting the sequencer is operating honestly.
Hard Finality: This is the point at which a transaction becomes truly irreversible, secured by the full consensus of the underlying settlement layer. This occurs only after the batch containing the transaction and its ZK proof has been posted and verified on the L1. This process is far more secure but also significantly slower, taking from a few seconds for rollups on Solana to ~13 minutes on Ethereum or even hours for complex L3s like Paradex.
This duality has direct consequences. Traders can and should rely on soft finality for low-risk, time-sensitive actions like placing or canceling orders. However, for high-value, security-critical actions like depositing or withdrawing large amounts of capital, users must wait for hard finality.
Composability
A primary and valid criticism of the Layer 2 rollup model is the sacrifice of atomic composability—the ability to interact with multiple independent protocols within a single, all-or-nothing transaction. When a CLOB moves to its own rollup, it becomes an "execution silo," unable to atomically compose with protocols on the L1. This fragmentation is a significant drawback for a generalized DeFi ecosystem built on "money legos".
However, proponents of specialized rollups make a compelling counterargument, particularly for perpetual futures markets. The dominant thesis, validated by the immense success of isolated platforms like Binance and Hyperliquid, is that perpetuals exchanges do not require deep, synchronous composability to thrive. They can and do function as highly successful "vertically integrated platforms". The calculus may be different for spot markets, which, as the Kuru team argues, are more foundational to a general-purpose DeFi ecosystem and "cannot" live in isolation.
The modular, ZK-powered rollup approach—combining high-performance off-chain execution, verifiable zero-knowledge proofs, and scalable data availability layers—solves fundamental issues that have historically plagued on-chain trading.
Hyperliquid's explosive success validated this model, simultaneously highlighting the inherent limitations of monolithic architectures like Solana. Meanwhile, specialized rollups like Bullet, Hibachi, and Lighter have demonstrated superior security guarantees, robust escape hatch mechanisms, and dramatically improved performance metrics.
Yet significant challenges remain: bootstrapping deep liquidity, simplifying the user experience, and creating transparent, fair forced-exit mechanisms are critical hurdles that must be overcome. These are not mere technical details but foundational requirements for mass adoption.
Looking forward, several trends will shape the future of the space:
Decentralizing Sequencers: Achieving performant yet genuinely decentralized sequencing will be essential to securing trust and long-term viability.
Cross-Chain Liquidity: Platforms that seamlessly aggregate liquidity across multiple chains will dominate the increasingly fragmented multi-chain landscape.
Privacy and Sovereignty: Privacy-enhanced platforms utilizing ZK technology will capture significant institutional and sophisticated retail market segments.
Blurring of L1/L2 Boundaries: The distinction between layers will continue to fade, with the most successful platforms defined by their combined performance, security, and user experience rather than their technical labels.
Ultimately, the CLOB Wars will not be won solely by superior technology. Victory will belong to platforms that synthesize the best aspects of centralized exchanges—deep liquidity, intuitive user experience, robust risk management—with the decentralization, transparency, and composability that only blockchain-based systems can offer. The race is fierce, innovation is rapid, and the ultimate winner is clear: the users, who stand to benefit from an open, performant, and truly decentralized financial system.
Share Dialog
Share Dialog
No comments yet