Blockchain Moves Toward 'Chain-Cloud' as Gas Abstraction Fades

Article by Fugui

The phrase “Gas is becoming obsolete” does not mean that Gas is about to disappear, but rather that its explanatory power as a unified resource abstraction for blockchains is diminishing. This assessment does not stem from a technological upgrade on a single chain, but rather from four entirely different directions, all pointing toward the same outcome.

The first direction comes from Hyperliquid. Its core users, who interact daily with a decentralized perpetuals exchange, see maker and taker trading fees rather than gas multiplied by gas price. HyperEVM still uses HYPE as gas, and both the base fee and priority fee from EIP-1559 are operational; however, the transaction layer has bundled the underlying resource costs into a service fee. When a user opens a long position on BTC, they are purchasing matching and settlement—not computational instructions within block space. Gas hasn’t been eliminated here; it’s simply been concealed within the pricing of the service.

The second direction comes from Solana. SIMD-0553 previously reached accepted and merged status, but its accompanying governance proposal, SGP-0003, was rejected in August 2026, and the resource fee model has yet to be implemented on mainnet. Nevertheless, its proposed separation—between a Base Inclusion Fee and a Resource Fee based on requested cost units—clearly indicates that Solana is pricing “inclusion” and “resource consumption” separately. The Base Inclusion Fee answers the question: “Is this transaction worth including in a block?” The Resource Fee answers: “How much network resource does this transaction actually consume?” The Resource Fee is grounded in requested cost units, which explicitly list signature cost, write lock cost, instruction data cost, program execution cost, and loaded account data cost. Solana itself is breaking Gas back down into the specific resources it originally obscured.

The third direction comes from ICP. The Internet Computer has always used cycles instead of gas; cycles are pegged to XDR and measure real resources such as compute, storage, bandwidth, and messaging. With the upcoming Cloud Engines in 2026, users will be able to select nodes, node providers, geographic locations, and replication scales, allowing the protocol to assemble a customized execution environment. The official term for this is a "user-owned slice of Internet Computer capacity." When users purchase a computing environment rather than a single transaction, the concept of gas itself recedes into the background.

The fourth direction is the most noteworthy, as it comes from Ethereum itself. On August 17, 2026, Ethereum launched the Platåberget testnet as an early public testing environment for the Glamsterdam upgrade. The official announcement warned in advance: wallets, indexers, and gas estimators relying on fixed gas limit assumptions may be broken by this upgrade. Glamsterdam adjusts the gas costs for operations such as state access; EIP-8037 and EIP-8038 respectively increase and partially decouple the gas costs for state creation and state access, making gas fees more closely reflect the actual work performed by nodes. EIP-7999 is advancing a unified multidimensional fee market. Ethereum itself is also breaking gas back down into what it originally concealed: state access, computation, data, and different types of node work.

Putting these four paths together makes the conclusion clear: Gas is retreating from being a core abstraction of the blockchain and reverting to a foundational settlement mechanism in the resource market.

Since Gas is being decomposed, the entire industry’s focus on the VM over the past decade may have been misguided from the start.

EVM is the most powerful smart contract execution environment in the ecosystem, and the network effects formed by Solidity, OpenZeppelin, Foundry, MetaMask, Uniswap, and Aave still have no rival. However, as the underlying execution model of a "world computer," EVM is increasingly struggling to bear the full responsibility of serving as the next-generation global verifiable computing infrastructure. The issue is not that it cannot run programs, but that it conflates execution, state, resources, and proof within a single abstraction.

From an execution model perspective, the EVM is a stack machine where each stack item is 256 bits. Modern CPUs prefer register-based architectures, and every ADD and MUL operation in the EVM involves significant stack manipulation. FuelVM uses 64-bit registers, directly addressing this historical overhead.

From a parallel processing perspective, EVM transactions do not declare state dependencies, making it impossible for the execution layer to schedule them in advance. Whether two transactions access the same state is nearly invisible within the EVM. In contrast, Solana’s Sealevel makes state dependencies visible to the scheduler—transactions explicitly declare which accounts they need to access, allowing the scheduler to analyze dependencies and execute transactions in parallel. This approach closely mirrors modern CPUs’ dependency graphs. Solana’s true innovation lies not in the sBPF instruction set, but in elevating “who can run in parallel” from a hidden runtime detail to a first-class citizen of the execution model.

From a proof model perspective, the EVM’s 256-bit stack structure and state access methods are inherently unsuited for ZK proving. zkEVM has long suffered from extremely high proving costs, and future blockchains must assume that execution is followed by proof. RISC-V is becoming increasingly important precisely because it is the first to give blockchains a low-level execution standard akin to a CPU—enabling integration with the LLVM and GCC ecosystems, register architecture, and standardized zkVMs. Ethereum and Polkadot both moving toward RISC-V is no coincidence.

Move’s contribution is on an entirely different level. Instead of pursuing “faster opcodes,” it transforms assets from data within smart contracts into language-level resources. In Solidity, double spending is prevented through audits and libraries; Move directly enforces this constraint through its language and type system. Resource safety, ownership, and abilities address not execution speed, but secure resource management.

Therefore, the competition among VMs is itself a false dilemma. EVM provides composable general-purpose smart contracts, SVM offers explicit dependencies and parallel scheduling, Move provides resource safety, RISC-V offers a general-purpose execution ISA, and zkVM enables provable execution. In the future, no single VM will win alone—it is the entire computer that will be reimagined.

Blockchain is not the first to invent resource pricing. It has simply taken ten years to rediscover what the computer industry has known for decades: computation is not a single resource, but a set of resources.

Ethereum’s classic model uses a single-dimensional Gas, where the fee equals GasUsed multiplied by GasPrice. This model is extremely simple but has a fundamental economic flaw: different resources that cannot be substituted for one another are priced using a single number. A transaction with very high CPU usage but low storage usage, and another with low CPU usage but very high storage usage, might both be priced at 100 Gas. This price does not accurately reflect the true bottlenecks of nodes.

EOS was already managing three separate resources—CPU, NET, and RAM—as early as 2018. RAM is a property right that can be purchased, used, and sold back, with its price dynamically adjusted via the Bancor algorithm. CPU and NET are leased resources, later evolving into the PowerUp model, which essentially involves renting computing and network resources for a specified period. EOS distinguished between Ephemeral Resources and Persistent Resources long before Sui separated Computation and Storage several years later.

TRON takes a different approach: Bandwidth corresponds to transaction bytes, Energy to contract execution, and TRON Power to governance. Users obtain resource allocations by staking TRX and can delegate these resources to others. TRON even allows DApp developers to set a "consume user resource percent," directly covering the Energy costs for users—this is already very close to the later concepts of Paymaster and Sponsored Transactions.

Hive has completely eliminated Gas, instead adopting a conversion from HP to RC. Resource Credits are not tokens, but renewable resource credits allocated to accounts based on their Hive Power. RC is automatically restored after consumption, with a recovery cycle of five days. This is essentially a quota model, well-suited for social and content-based applications, but not ideal for complex DeFi or high-frequency computations.

These older chains addressed "resource measurement" but not "resource markets" or "resource abstraction." Users still need to understand concepts like CPU, NET, RAM, Energy, and Bandwidth to use the chain effectively—this is one reason they ultimately failed to become the standard solution for user experience.

But their underlying ideas haven't disappeared. Solana separates computation into CUs, turning transactions into resource requests. Sui splits fees into Computation and Storage, explicitly recognizing for the first time in a mainstream L1 that "executing once" and "permanently storing 1MB" are two entirely distinct economic actions. Polkadot uses ref_time and proof_size to form a two-dimensional Weight, combined with storage deposit to handle long-term state growth. Moving from Gas to multidimensional resources isn't about newer chains being smarter than older ones—it's about the entire industry gradually acknowledging that blockchain nodes consume far more than just one type of real resource.

Resource measurement is only the first step. The real questions to address are: Who sets the price for these resources? How are resources allocated? Can resources form a market?

Hedera offers a compelling answer worth studying. It breaks down transaction costs into Node Fee, Network Fee, and Service Fee, weighted behind the scenes by Bandwidth, Processing, Storage, and Duration—assigning higher prices to scarcer resources. This is no longer Opcode Gas Accounting, but Resource Cost Accounting. More interestingly, in 2026, Hedera introduced Simple Fees, packaging complex resource rates into a Base plus Extras format. The underlying system remains multidimensional, while the user interface is unidimensional—an engineering approach that is entirely feasible.

ICP has taken another step forward. Cycles are pegged to XDR, so a surge in token price won’t cause server resource costs to rise proportionally—developers face relatively stable fiat-denominated resource prices. Compute Allocation is billed directly based on core usage proportion, with a fixed number of cycles charged per second for one percent of a core. This is not a transaction fee; it’s a cloud computing bill.

Filecoin demonstrates the purest form of a resource market. Clients and storage providers negotiate prices directly, and transactions are published on-chain to form the Storage Market. Prices are not fixed by the protocol but are determined directly by supply and demand. This is the closest example in the industry to a decentralized resource market.

In addition, block space includes a frequently overlooked resource dimension: inclusion priority and ordering rights. Priority fees, Jito-style auctions, and execution tickets are essentially pricing mechanisms for determining who gets included first and in what order transactions are executed. While compute, storage, and data availability correspond to the physical and verification costs of nodes, priority corresponds to the scarce positional slots within a block. MEV can be seen as a crude and fragmented pricing of this special resource; a mature resource market should explicitly incorporate it rather than leaving it长期 outside the protocol.

From these practices, four levels can be abstracted: Resource Accounting addresses "how much was consumed," Resource Pricing addresses "how much it costs," Resource Allocation addresses "who receives the resources," and Resource Market addresses "how resources are freely traded." The future of blockchain pricing will not be "better gas," but a multidimensional resource pricing system. Fees equal the sum of all resource usage multiplied by their respective prices, where resources include Compute, State, Storage, Bandwidth, DA, Proof, Concurrency, Messaging, and Priority.

The World Computer has never been like a computer—not because it lacks a fast enough VM, but because it only has “execution,” without comprehensive resource management, scheduling, pricing, isolation, and markets. A real computer doesn’t let every program define its own CPU price. A real cloud doesn’t require users to understand “how much RAM I need to buy right now.”

ICP Cloud Engines represent the first true productization of "blockchain plus cloud." Users select nodes, node providers, geographic locations, and replication scales, and the protocol assembles a customized execution environment. The official dashboard already displays data centers, node providers, and machine specifications. However, it's important to temper enthusiasm—this does not automatically prove that ICP is the final answer. What it offers is a promising early prototype of blockchain-based cloud: blockchain can organize nodes, hardware, locations, replication, and execution environments into a cloud service.

Hyperliquid and GameFi demonstrate another direction. Hyperliquid users purchase trading services, while GameFi players purchase gaming experiences. Immutable’s Orderbook allows users to sign listings for NFTs with zero gas fees, separating the gas payer from the transaction signer. Gas costs shift from user expenses to platform operational costs—aligning perfectly with Web2 economic models: users pay for the product, not the infrastructure.

ChainCloud is not AWS plus a token. It is a globally verifiable pool of computing resources managed by protocol, priced by the market, provided by nodes, proven by cryptography, and connected by unified execution standards.

Pushing this logic to its conclusion, the structure of the next-generation blockchain is no longer a "single chain," but a five-layer stack.

At the top is the Service Layer. Applications such as Trading, Game, Storage, and AI directly serve users, who see business prices, not resource prices.

Below is Resource Abstraction. This layer is responsible for packaging underlying multidimensional resources into service interfaces that upper layers can understand. Ethereum's EIP-7999 has already designed in this direction: the underlying layer is multidimensional, while the user interface is unidimensional—users only need to provide a single unified max fee, and the protocol internally allocates the budget across different resources.

Below is the Resource Market. Compute, State, Storage, DA, Bandwidth, Proof, and Priority each have their own prices, determined by supply and demand, forming tradable resource pools.

Below the Resource Market is the Parallel Runtime, which handles dependency graph analysis, concurrent scheduling, and state access management. Sealevel, Block-STM, and Sui Object Model all belong to this layer.

At the lowest level are the Execution ISA and Distributed State. RISC-V is present here as just one general-purpose execution standard for the chain cloud, not the core. The true core is the combination of ISA, Runtime, State, Resource Market, and Proof.

Users should never see terms like Gas, Gwei, CU, ref_time, proof_size, or storage deposit. They should only see phrases like “Exchange 100 USDC for ETH,” “Buy a sword,” “Deposit 1GB to store for 100 years,” or “Open a long position on BTC.”

The underlying pricing formula is multidimensional: Fee equals the usage of each resource multiplied by its current price. But the user-layer pricing formula is entirely different: User Price equals Service Value plus Resource Cost, plus Risk Premium, plus Platform Margin. The underlying layer increasingly precisely measures resources, while the top layer increasingly fully hides them. This is what a truly mature computer should look like.

Gas is not death, but abdication. It has transformed from the resource itself into a packaging layer for the resource market.

The evolution path of blockchain has become increasingly clear: from Blockchain to World Computer, to Resource Computer, to Resource Market, then to Decentralized Cloud, and ultimately toward Chain-Cloud.

Blockchain solves trust; the cloud solves resources. The mission of the next-generation blockchain is not to move all computation onto the chain, but to transform global computing resources into a verifiable, tradable, and programmable public infrastructure. When this transformation is complete, blockchain will no longer be just a faster chain, but a truly cryptographic cloud.