For a dApp that needs to write a large amount of data on-chain, which token's network is the cheapest for data storage?

If you mean storing arbitrary application data directly on a blockchain/state layer, rather than merely storing a hash/pointer, Sui (SUI) is one of the strongest candidates for low-cost native on-chain storage.

• Sui: its storage fee is currently 76 MIST per storage unit, with 100 storage units per byte. That works out to 0.0076 SUI per KB under the documented rate, paid upfront for persistent storage. Deleting data can return 99% of the storage fee under the cited model.Sui+1
• Solana (SOL): uses a refundable rent-exemption deposit for account data. Solana has been rolling out a 90% reduction in the storage/rent parameter, making it substantially cheaper than its historical pricing, but the exact cost depends on the current rollout stage and account structure.Solana+1
• Arweave (AR): is different: it's specifically designed for permanent data storage, rather than general-purpose blockchain state. It charges a one-time storage fee; its current calculator gives live pricing.Arweave+1
• Filecoin (FIL): is generally better thought of as decentralized storage rather than putting the data into every validator's blockchain state. Filecoin Onchain Cloud currently quotes $2.50/TiB/month per copy plus a per-dataset proving fee.Filecoin Onchain Cloud Documentation

The important distinction

For a dApp storing a large amount of data, I'd separate the choices like this:

So if your requirement is literally “I need to write tens/hundreds of GB of application data on-chain and have the blockchain itself maintain it,” Sui is particularly interesting. Its storage-fund design explicitly charges the writer for long-term storage rather than making future validators/users subsidize historical data.GitHub+1

However, “cheapest” in USD is highly dependent on the token price and the network's current storage parameters. If you tell me roughly how much data you need to store (e.g. 10 GB, 1 TB, 100 TB) and whether it needs to be mutable, I can calculate a current SUI vs SOL vs AR vs FIL cost comparison.

If by “write a large amount of data on-chain” you mean the data itself must be stored and retrievable from the blockchain/network—not merely a hash or pointer—then the answer is different from ordinary transaction-fee comparisons.

Short answer

Filecoin (FIL) is generally among the cheapest decentralized options for large-scale storage, but its data is not literally stored inside every Filecoin blockchain block. Filecoin uses the blockchain to coordinate storage deals and proofs, while the bulk data is held by storage providers.Filecoin+1

For permanent, immutable data, Arweave (AR) is the more relevant comparison. You pay upfront for long-term/permanent storage, but it is substantially more expensive per GB than Filecoin's current storage pricing.0G+1

So I'd break it down like this:

For perspective, a recent 2026 comparison puts Filecoin cold archival around $0.19/TB/month, versus roughly $5–8/GB for Arweave permanent storage—though these figures fluctuate and aren't directly equivalent because the retention models differ.0G

But there's an important distinction for your dApp

If you mean:

“I need my smart contract to store, say, 100 GB–10 TB of application data, and users need to interact with that data trustlessly.” I would not put the raw data directly into conventional L1 smart-contract storage. It becomes enormously more expensive and can create state-bloat problems.

A much more economical architecture is:

Blockchain → hash/commitment + metadata → decentralized storage/DA layer → actual data

For example:

Smart contract
    │
    ├── content hash
    ├── dataset ID
    ├── ownership/access rules
    └── storage pointer
             │
             ▼
       Filecoin / Arweave / DA layer
             │
             ▼
          Large data

If you tell me roughly how much data you're talking about (e.g. 1 GB, 100 GB, 10 TB), whether it needs to be permanent, and whether smart contracts need to read individual records directly, I can calculate the approximate $/GB or $/TB for the major candidates (including SUI, Solana, Avalanche, Arweave, Filecoin, Celestia, and Ethereum L2s) and identify the economics that fit your dApp.

When building a decentralized application (dApp) that needs to write a large amount of data , writing it directly "on-chain" into smart contract state (like Ethereum, Solana, or BNB Chain) is economically and architecturally prohibitive. Traditional L1/L2 gas costs scale poorly for bulk data, making Decentralized Storage Networks (DSNs) paired with smart contract anchors the correct architectural pattern.

The most cost-effective and practical token networks for large-scale data storage depend on whether you need temporary/lease-based storage, high-throughput modular data availability (DA) , or one-time permanent storage.0g.ai

Comparison of the Cheapest Data Storage Networks

1. Filecoin (FIL)
   • Best For: Cheapest bulk, cold archival, and long-term leasing storage.
   • Cost Structure: Pay-as-you-go leasing model via storage providers.
   • Approximate Cost: ~$0.19 per TB/month (varies by market supply, demand, and deal parameters).
   • Trade-off: Retrieval can have higher latency (~45 seconds), and you must manage periodic renewal deals to avoid data expiration.binance.com+3
2. Walrus / Sui Network (WAL / SUI)
   • Best For: Highly efficient, cost-optimized dynamic/blob storage using erasure coding.
   • Cost Structure: Powered by the W A L cap W cap A cap L𝑊𝐴𝐿 token on the S u i cap S u i𝑆𝑢𝑖 ecosystem with built-in fiat-cost-smoothing mechanisms.
   • Approximate Cost: Extremely competitive (significantly lower redundancy overhead than traditional replication models, running around a fraction of older DSN costs).
   • Trade-off: Newer ecosystem compared to Filecoin or Arweave, but built explicitly for efficient binary large objects (blobs).bit2me.com+20G Storage (Zero Gravity)
   • Best For: High-throughput, programmable on-chain/decentralized storage designed for AI datasets and high-frequency dApps.
3. Arweave (AR)
   • Best For: Permanent, immutable storage paid once.
   • Cost Structure: One-time upfront endowment fee that assumes storage costs will decrease over time.
   • Approximate Cost: Roughly $5 to $8 per GB (one-time fee for lifetime storage).
   • Trade-off: Prohibitively expensive if you are writing terabytes of transient or frequently changing data, as you pay for permanence upfront.binance.com+2

Summary Recommendation

• If your dApp requires the absolute lowest recurring cash burn for massive historical or cold data, use Filecoin (F I L cap F cap I cap L𝑭𝑰𝑳).
• If your dApp needs a modern, fast ecosystem optimized for efficient blob handling and fast integration, look at Walrus (W A L cap W cap A cap L𝑾𝑨𝑳) on Sui.
• If your dApp requires high-speed read/write throughput (e.g., streaming or active application data), look into 0G Storage.digitalsurge.com.au+2

(Note: Never store raw large files directly in EVM/Solana contract storage. Instead, store the heavy data on one of the networks above and save only the cryptographic content hash/URI on-chain via a low-cost network like Base, Arbitrum, or Solana).

To help narrow down the exact network, tell me:

• What is the estimated total size of the data (e.g., gigabytes vs. terabytes)?
• Does the data need to be permanently immutable , or will it be updated/deleted over time?
• Is your dApp deployed on a specific L1 or L2 already (like Ethereum, Solana, or Sui)?

If a dApp needs to store a truly massive/large amount of data , storing it directly "on-chain" in traditional smart contract storage (like Ethereum state storage) is financially prohibitive. Instead, developers choose between Modular Availability Layers (if the data needs to be verified by smart contracts for a short/medium window) or Decentralized Storage Networks (for permanent or long-term archiving).

The cheapest networks depend entirely on your retention requirements and whether you need execution logic attached:

1. For Temporary/Modular Data Availability (Best for rollups & high-throughput dApps)

If your dApp just needs data published so that smart contracts can trust it and verify it before pruning it later, Data Availability (DA) layers are the cheapest and most efficient.

• Celestia (TIA) / Avail / EigenDA: These networks decouple data posting from execution. Celestia uses optimal data availability sampling, making it dramatically cheaper to post large batches of rollup/dApp data compared to storing data in L1 smart contract state.
• Ethereum L2 Blobspace (Base, Arbitrum, Optimism via EIP-4844): Posting data via blobs on L2s is significantly cheaper than standard Ethereum calldata, though still more expensive per gigabyte than specialized DA layers or permanent storage networks.

2. For Permanent Storage (Pay once, store forever)

• Arweave (AR): Instead of a recurring subscription, Arweave charges a one-time upfront fee based on the long-term decline of storage hardware costs. Storing large static files (megabytes to terabytes of historical data, media, or states) becomes a flat upfront cost that covers centuries of replication. While the initial lump sum might look high compared to a monthly lease, it is mathematically the cheapest long-term permanent solution because you never pay again.binance.com+1

3. For Rent-Based/Leased Decentralized Storage (Cheapest short-to-medium term per GB)

If you do not need permanent storage baked into a crypto endowment, leased decentralized storage networks offer low recurring costs per gigabyte:

• Filecoin (FIL): Operates on a market-based leasing model where you pay storage providers over time. It is exceptionally cheap for massive bulk data (terabytes of unstructured data), often costing a fraction of Arweave's permanent endowment upfront, but requires ongoing renewal transactions.binance.com+2
• Walrus (by Sui) / Storj / Sia: Newer alternatives leverage advanced erasure coding rather than heavy replication, driving down active storage leasing costs even further than classic Filecoin models.binance.com+2

Summary Recommendation

• Choose Celestia or an Ethereum L2 blob space if your smart contracts need immediate, trustless access to the data stream for validation before it gets archived.
• Choose Filecoin (or emerging networks like Walrus ) if you have raw, bulky data and prefer a low-cost, recurring rental model.
• Choose Arweave if you need the data to be permanently immutable with a single one-time fee.binance.com+1

To help narrow down the exact network, tell me:

• How much data (in megabytes or gigabytes) do you expect to write per day/month?
• Does the data need to be permanently stored , or just temporarily available for smart contract verification?
• Do you need on-chain smart contracts to read this data natively?

If your decentralized application (dApp) needs to store a truly large amount of data (megabytes to gigabytes), storing it directly into standard smart contract state storage (like Ethereum storage slots) on any general-purpose Layer 1 or Layer 2 is economically unfeasible.

Instead, the "cheapest" network depends heavily on whether you need Data Availability (DA) / Rollup-style data or Permanent/File Storage (DePIN).

1. For Dedicated Data Availability (DA) / High-Throughput Appchains: Celestia (TIA)

If your dApp is structured as a rollup or needs verifiable, high-throughput blob data posted to a consensus layer rather than traditional database files:

• The Token/Network: Celestia (T I A cap T cap I cap A𝑇𝐼𝐴).
• Cost Efficiency: Celestia pioneered modular Data Availability and remains significantly cheaper than traditional execution-layer storage or even Ethereum’s EIP-4844 blobs (often roughly 5x to 10x cheaper than Ethereum blobs, and orders of magnitude cheaper than old-school calldata).blockeden.xyz+1
• Cost Range: Historically averages a fraction of a cent per megabyte (fractions of a cent to ~$0.07–$0.70 per MB depending on network congestion and T I A cap T cap I cap A𝑇𝐼𝐴 token pricing).blockeden.xyz+1
• Best for: Rollups, high-frequency gaming states, or app-chains needing proof that data was published without bloating an EVM state.

2. For Permanent, Immutable Raw Data (Files, JSON, Media): Arweave (AR)

If you need raw data (not execution state) stored forever with a single upfront payment:

• The Token/Network: Arweave (A R cap A cap R𝐴𝑅).
• Cost Efficiency: Arweave uses a permanent storage model—you pay a one-time fee calculated based on the long-term decline of hard drive costs.binance.com+1
• Cost Range: Varies based on the A R cap A cap R𝐴𝑅 token price and storage endowment calculations, but effectively guarantees multi-hundred-year persistence for a fixed one-time fee per gigabyte.
• Best for: NFT assets, historical permanent archives, heavy frontend hosting, or permanent decentralized databases (via tools like Bundlr/Irys).

3. For Cheap Leased/Rented Decentralized Storage (DePIN): Filecoin (FIL) or Walrus

If you have massive quantities of data (gigabytes to terabytes) that you don't necessarily need permanently endowed upfront, but want cryptographically verified cloud storage:

• The Token/Network: Filecoin (F I L cap F cap I cap L𝐹𝐼𝐿) or emerging alternatives like Walrus (Sui ecosystem).
• Cost Efficiency: Filecoin uses a rental/leasing model where storage providers bid on your storage deals. It is dramatically cheaper per gigabyte-month than permanent storage, though you have to renew or set up long-term deals. Emerging erasure-coded networks like Walrus boast even lower redundancy overhead costs.
• Best for: Large datasets, video streaming, AI training data, or cold/warm data backup.binance.com+1

Summary Recommendation

• Do not use standard general-purpose L1s/L2s (like Ethereum, Solana, or Polygon main state) for bulk raw data; you will pay high state-rent or transaction bloat fees.
• Choose Celestia (T I A cap T cap I cap A𝑻𝑰𝑨) if you need low-cost cryptographic data availability for an active transaction/rollup stream.
• Choose Arweave (A R cap A cap R𝑨𝑹) if you want "write once, store forever" immutability.
• Choose Filecoin (F I L cap F cap I cap L𝑭𝑰𝑳) or decentralized cloud alternatives if you are handling massive, traditional file-blobs on a budget via leasing.blockeden.xyz+2

If you can share:

• The approximate size of the data (e.g., kilobits vs. gigabytes per day)
• Whether the data needs to be read/processed by a smart contract natively, or just referenced via hash

I can help you pinpoint the exact protocol and architectural pattern to use.