Modular Blockchains: The Architecture Reshaping Crypto

The Limits of the Original Design

When Satoshi Nakamoto published the Bitcoin whitepaper in 2008, the architecture was deliberately monolithic. A single chain would handle everything: verifying transactions, reaching agreement among participants, publishing data, and anchoring finality. The elegance was in the simplicity. One system, one security model, one ledger. For a payments network operating at modest throughput, this was sufficient.

Ethereum extended the model. It kept the same integrated architecture but added programmability through the Ethereum Virtual Machine, enabling smart contracts, decentralized applications, and eventually the infrastructure backbone for a multi-trillion-dollar financial ecosystem. Yet as demand grew, so did the contradictions baked into the design. The Ethereum network could process roughly 15 transactions per second. During the DeFi summer of 2020 and the NFT boom of 2021, average gas fees regularly exceeded $50 per transaction — at peak congestion, fees for complex contract interactions surpassed $200. Institutional adoption at scale was structurally impossible under those conditions.

The problem was not Ethereum specifically. It was the monolithic model itself. Every node on the network was required to perform every function simultaneously: compute every transaction, verify every state change, store all historical data, and participate in consensus. Scaling meant either increasing block sizes — which raised hardware requirements and threatened decentralization — or finding a fundamentally different approach. The industry chose the latter.

Disaggregating the Stack

The modular blockchain thesis is built on a deceptively simple insight: the four core functions of a blockchain network do not need to be performed by the same layer. Execution, settlement, data availability, and consensus can each be handled by specialized systems optimized for their specific role. The result is a horizontally scalable architecture where each layer improves independently without forcing trade-offs onto the others.

Execution: Where Computation Lives

Execution is the computational layer — the environment in which smart contracts run, state transitions are calculated, and user-facing applications operate. In a monolithic system, execution happens on the base chain alongside everything else. In a modular design, execution is deliberately moved off the base layer entirely. Rollups — both optimistic and zero-knowledge variants — are the dominant execution environments in the current architecture. Arbitrum and Optimism collectively process more transactions daily than Ethereum's base layer, yet they do so at a fraction of the cost by batching execution off-chain and submitting compressed proofs or transaction data back to Ethereum for finalization.

Settlement: The Anchor of Finality

Settlement is where finality lives. It is the function that makes a transaction irreversible — the layer that resolves disputes, enforces state correctness, and anchors the canonical chain. Ethereum's Layer 1 currently serves as the settlement layer for the broader rollup ecosystem. When Arbitrum posts a batch of transactions back to Ethereum, it is not re-executing those transactions on-chain; it is submitting a cryptographic commitment to their correctness and relying on Ethereum's security model to make that commitment permanent. The settlement layer does not need to be fast. It needs to be trustworthy.

Data Availability: The Silent Prerequisite

Data availability is the least intuitive of the four functions, yet arguably the most critical for the long-term security of the modular stack. DA does not mean that data is permanently stored — it means that transaction data is published and accessible at the moment it is produced, so that validators, full nodes, and proof systems can verify correctness and reconstruct state if needed. Without guaranteed data availability, fraud proofs become unverifiable, validity proofs lose their grounding, and the trust model of rollups collapses entirely. This function has historically been bundled with settlement on Ethereum's base layer, but it is now becoming its own specialized market.

Consensus: Coordinating Agreement

Consensus is the mechanism by which validators agree on the ordering and inclusion of transactions. It enforces economic security through staking, slashing, and validator incentives. In a modular design, consensus may be inherited from the base layer, provided by a dedicated validator set, or delegated entirely to an economic security marketplace. EigenLayer, which had accumulated over $15 billion in restaked ETH at its peak, is building precisely this: a shared security infrastructure where new protocols can rent Ethereum's validator set rather than bootstrapping their own from scratch.

The Rollup-Centric World

Ethereum's current roadmap is explicitly organized around what its developers call the rollup-centric thesis: the base layer should become a highly secure, highly decentralized settlement and data availability substrate, while execution scales through a proliferating ecosystem of rollups. This is not a stopgap measure. It is Ethereum's long-term architectural identity.

The two dominant rollup paradigms — optimistic and zero-knowledge — approach the problem of trustless execution from different directions. Optimistic rollups, exemplified by Arbitrum One and Optimism's OP Stack, assume transactions are valid by default and allow a challenge period — typically seven days — during which any party can submit a fraud proof disputing a posted state root. The model is pragmatic: fraud proofs are relatively simple to implement, and the seven-day window has not historically impeded capital flows in meaningful ways for most use cases. Arbitrum has routinely hosted over $10 billion in total value locked, a figure that speaks to market confidence in the model despite its theoretical limitations.

Zero-knowledge rollups take a stricter approach. Rather than optimistic assumptions, ZK rollups — including zkSync Era, StarkNet, Polygon zkEVM, and Scroll — generate cryptographic validity proofs that mathematically guarantee the correctness of every state transition. There is no challenge window because there is nothing to challenge; the proof is self-verifying. Finality is near-instant. The trade-off has historically been computational overhead: generating ZK proofs is expensive, and supporting the full complexity of the Ethereum Virtual Machine in a ZK-compatible environment required years of engineering work. That work is now largely complete, and ZK rollups are broadly considered the long-term direction for Ethereum's execution layer.

The Data Availability Market

As rollups have scaled, the cost and throughput limitations of posting data to Ethereum's base layer have become an active constraint. Ethereum's EIP-4844 upgrade, deployed in March 2024, introduced "blobs" — a new data format specifically designed for rollup data that dramatically reduced costs. Within weeks of launch, transaction fees on Arbitrum and Optimism fell by more than 80 percent, in some cases approaching fractions of a cent for standard transfers. The upgrade was a meaningful inflection point, but it was also a preview of the emerging competition in the DA layer.

Celestia launched its mainnet in late 2023 as the first purpose-built data availability network. Rather than executing transactions or enforcing settlement, Celestia does one thing: it ensures that transaction data is published and available, using a technique called data availability sampling to allow light nodes to verify DA without downloading entire blocks. The design enables extraordinarily high throughput for data publication at a fraction of the cost of Ethereum blobs. EigenDA, built on EigenLayer's restaking infrastructure, takes a different approach — leveraging Ethereum's existing validator set to provide DA guarantees without a separate consensus mechanism. Avail, spun out of Polygon, offers a third variant. The competition among DA providers is now genuine, and the pricing pressure it creates flows directly to rollup operators and, ultimately, to end users.

For investors, the emergence of a competitive DA market raises a structural question about value accrual. If DA becomes commoditized — if Celestia, EigenDA, and Avail drive fees toward marginal cost — then the fee revenue that once accrued entirely to Ethereum validators will be distributed across a broader set of infrastructure providers. The total addressable market for blockspace expands, but the margin per unit compresses. This dynamic is not unfamiliar to anyone who has studied the history of cloud computing or telecommunications infrastructure buildouts.

What Modularity Means for Capital Allocation

The modular thesis has direct implications for how sophisticated investors should think about protocol exposure. In a monolithic world, owning the base layer token was a reasonably clean bet on the entire stack — execution fees, settlement fees, and data costs all accrued to the same asset. Modularity disaggregates that value capture across potentially dozens of specialized layers, each with its own token, security model, and fee market.

Ethereum retains structural advantages in this environment. Its role as the dominant settlement layer — the chain that the broader ecosystem trusts for finality — is not easily replicated. The social consensus, validator decentralization, and institutional familiarity that underpin Ethereum's security model represent years of compounding network effects. Stakers who secure the settlement layer capture fees from every rollup that posts to Ethereum, regardless of which DA layer those rollups use. EIP-4844 reduced per-blob fees substantially, but it also dramatically increased blob volume, keeping aggregate revenue meaningful.

Layer 2 tokens represent a different risk and reward profile. They capture execution fees within their specific environment, and their value is tied to the economic activity — DeFi volume, NFT transactions, gaming interactions — generated within their rollup. The sequencer revenue models vary significantly: some rollups centralize sequencing and capture the spread between execution costs and DA costs as protocol revenue, while others are moving toward decentralized sequencer sets that distribute this revenue more broadly. Understanding the fee architecture of individual rollups is now a prerequisite for informed capital allocation in the space.

The DA layer tokens — Celestia's TIA being the most prominent — represent a bet on the commoditization of a critical infrastructure function. If modular chains proliferate and rollup data volumes grow by orders of magnitude, the aggregate fee revenue flowing to DA providers could be substantial even at low per-unit margins. The counterargument is that DA is precisely the kind of undifferentiated infrastructure where competition destroys pricing power. Historical analogies in technology infrastructure suggest the truth lies somewhere between those poles: dominant DA providers may capture durable margin through reliability, trust, and ecosystem lock-in, even as pricing remains competitive.

The Bottom Line

The shift from monolithic to modular blockchain architecture is one of the most consequential structural changes in the history of the industry. It resolves, at least in principle, the scalability trilemma that plagued first-generation networks by separating functions that were never technically required to coexist on a single layer. The result is a multi-layered stack in which execution environments can scale to consumer internet throughput, settlement layers can remain lean and trustworthy, and data availability becomes a commoditized input purchased at market rates.

For investors, the modular world is more complex to navigate than the simple base-layer thesis that characterized the 2017 and 2020 cycles. Value no longer accumulates neatly in a single asset. It distributes across settlement layers, execution environments, DA networks, and shared security infrastructure — each with distinct economics, competitive dynamics, and risk profiles. The protocols that will define the next decade are already operating in production. The question is not whether modularity will win. It already has. The question is which layers will prove most defensible as the architecture matures and competition intensifies across every level of the stack.