How Gas Fees Work (and Why They Exist)
On public blockchains like Ethereum, every computation costs gas. Gas is a unit that measures the computational effort required to execute an operation. Sending a simple token transfer might cost 21,000 gas. Executing a complex smart contract could cost millions. The actual price you pay depends on the gas price, denominated in gwei (billionths of ETH), which fluctuates based on network demand.
Gas fees exist for a practical reason: they prevent spam. On a network where anyone can submit transactions, without a cost mechanism, an attacker could flood the network with garbage transactions for free. Gas fees ensure that every computation has a price, which naturally limits abuse. They also compensate validators for the resources they contribute to securing the network.
The problem for enterprises is volatility. During the DeFi summer of 2020 and the NFT boom of 2021, Ethereum gas prices spiked to levels that made many applications economically unviable. A simple ERC-20 token transfer that cost $0.30 in a quiet period could cost $80 during peak congestion. Smart contract interactions were even worse, complex DeFi operations sometimes carried gas costs exceeding $500 per transaction.
Enterprise Cost Models: Predictable and Fixed
Enterprise IT budgets are built on predictability. You know what your cloud instances cost per month. You can forecast storage growth. Your database licensing has a defined price schedule. Even variable costs like API calls or data transfer have published rate cards.
Permissioned blockchain costs follow this familiar model. You run nodes on your own infrastructure or cloud instances, and the costs are the same as running any other distributed application, compute, storage, networking, and personnel. There are no gas fees because there is no public network to compensate. Transaction costs are effectively zero at the network level, though you bear infrastructure costs.
A Hyperledger Fabric deployment with three organizations, each running two peers and an orderer node, might cost $2,000 to $5,000 per month in cloud infrastructure per organization. That is predictable, budgetable, and comparable to running any other enterprise application. Whether you process one hundred transactions or one hundred thousand, the infrastructure cost is roughly the same.
Comparing Apples to Oranges
Direct cost comparisons between gas fees and enterprise infrastructure are misleading because the models pay for different things.
Gas fees pay for security you do not have to operate. When you pay gas on Ethereum, you are buying access to a network secured by billions of dollars in staked capital, operated by tens of thousands of validators worldwide. You do not manage those validators. You do not patch those servers. You do not negotiate uptime SLAs. The security is baked into the protocol and paid for per transaction.
Enterprise infrastructure costs pay for resources you control. You manage the servers, hire the engineers, and handle the operations. You get predictability and control, but you also carry the operational burden and the risk. If your three consortium partners decide to shut down their nodes, your network goes offline. On Ethereum, the network keeps running regardless of what any individual participant does.
A fair comparison considers total cost of ownership. A permissioned network with five participants might cost $200,000 per year in combined infrastructure and personnel across all parties. Anchoring the same volume of transactions to Ethereum via a layer-2 rollup might cost $10,000 to $30,000 per year in gas fees, but each participant still needs engineers who understand the technology, so personnel costs are comparable.
Layer-2 Changes the Math
The emergence of layer-2 solutions has significantly altered the public chain cost equation. Rollups, both optimistic (Arbitrum, Optimism) and zero-knowledge (zkSync, StarkNet), batch hundreds or thousands of transactions into a single on-chain proof. The gas cost is amortized across all transactions in the batch.
On Arbitrum, a typical transaction costs $0.01 to $0.10. On zkSync Era, costs are similar. For enterprise use cases with moderate transaction volumes, supply chain events, audit entries, credential verifications, these costs are negligible. A system processing 10,000 transactions per day on a layer-2 would cost roughly $100 to $1,000 per day in fees, which is competitive with permissioned infrastructure once you factor in operational savings.
Dedicated layer-2 chains, sometimes called app-chains or app-specific rollups, take this further. Platforms like Caldera or Conduit let enterprises deploy their own rollup chain with dedicated block space and predictable costs. You get the isolation and performance of a private network with the security guarantees of public chain settlement. Costs are typically $3,000 to $10,000 per month for the rollup infrastructure plus base-layer settlement fees.
Hidden Costs That Vendor Pitches Skip
Both models have costs that do not appear in simple comparison tables.
For permissioned networks: governance overhead. Coordinating software upgrades, handling disputes, managing membership changes, and maintaining consensus on data standards all require human time. In mature consortiums, governance meetings consume significant executive attention. This cost does not show up on an infrastructure bill, but it is very real.
For public networks: key management and wallet infrastructure. Enterprises need hardware security modules, multisig wallets, key rotation procedures, and disaster recovery for private keys. Losing a private key on a public chain can mean losing access to assets or contracts permanently. The infrastructure and processes around key management represent a cost category that does not exist in traditional enterprise IT.
For both: smart contract auditing. Before deploying business logic to any blockchain, the code needs to be audited by specialized security firms. Audit costs range from $10,000 for simple contracts to $500,000 or more for complex DeFi protocols. This is a recurring cost, every significant code change requires a re-audit.
Choosing Based on Economics
The economic decision framework is straightforward once you strip away the ideology. If your transaction volumes are high and cost predictability is critical, permissioned or dedicated layer-2 chains make sense. If your transaction volumes are low to moderate and you want to avoid operating infrastructure, layer-2 public chains are increasingly competitive. If you need the absolute strongest security guarantees and censorship resistance, Ethereum’s base layer is unmatched but expensive.
Most enterprise deployments end up somewhere in the middle, running business logic on a permissioned or layer-2 chain and anchoring critical proofs to a public chain. That hybrid model optimizes for cost, performance, and trust simultaneously. The projects that get into trouble are the ones that make the cost decision based on ideology rather than arithmetic.