SQD Network

SQD Network, formerly Subsquid, is a decentralized blockchain data platform: not the TCL "SQD" TV panel or the Aeroplan status metric that share the acronym. Developers and enterprises use its Portal API to stream raw and decoded onchain data, real-time and historical, from Solana and 120+ other networks through one HTTP endpoint. Behind Portal sits a permissionless network of 2,000+ worker nodes holding more than two petabytes of validated blockchain history, per its website in October 2026.

What sets SQD apart from an RPC provider is that each block is cross-checked before it is served and every worker response is signed, so data errors are detectable and slashable. Subsquid Labs GmbH, based in Zug, Switzerland, was acquired by Nasdaq-listed Rezolve Ai in October 2025, and the SQD token continues to run the network.

How SQD Network works: Portal, worker nodes and data validation

The network separates ingesting data from serving it. In the current bootstrapping phase, Subsquid Labs is the main data provider: it pulls raw blocks from several independent RPC endpoints, checks each one, and compresses validated data into Parquet chunks. A scheduler spreads those chunks across worker nodes, which store them locally and run queries with DuckDB, an in-process analytical database.

When a consumer queries Portal, the request goes to whichever workers hold the relevant data. Each worker signs its response with its private key, which gives consumers a verifiable record of who returned what. SQD's site quotes a 48.5ms P90 and 27ms median Portal response time.

How SQD validates blockchain data

SQD says it runs six checks per block before data reaches a client. The checks it documents include:

• Transaction and receipt trie roots: the ingester rebuilds the Merkle Patricia Trie root from the transactions it received and compares it with the block header, so a missing or altered transaction shows up as a mismatch.
• Logs bloom filter: the 2,048-bit bloom filter is recomputed from ingested logs and compared byte for byte.
• Log index continuity: log indices within a block must run unbroken from zero.
• Parent-hash chaining: each header commits to its predecessor, so chain continuity can be checked from one honest anchor.
• Multi-RPC arbitration: disagreements between endpoints are resolved explicitly rather than taking the first answer.

Several of these are EVM-specific; Solana data goes through the same multi-source ingestion and signed-response model. A second system, Snoopy (in beta), watches whether different workers return the same result for the same query. DuckDB queries over the same Parquet data should never diverge, so a divergence is evidence of a faulty or dishonest worker. Snoopy produces fraud proofs (Groth16 proofs generated with the SP1 zkVM) and submits them onchain to slash the offending worker's bond.

The difference matters in practice. In a benchmark Morpho ran on its production Portal over a rolling 90-day window in 2026, SQD returned roughly 100% correct data against 85-93% for the four RPC tiers tested, with 99.98% query success across 33.9 million queries and up to 41x faster batch reads than a premium RPC. That is a customer benchmark published by SQD, not an independent test.

SQD Portal for Solana: instructions, token balances and rewards

Solana is a first-class chain on Portal. The Solana dataset exposes six tables: instructions, transactions, balances, token balances, rewards and logs. Mainnet is served as a full archive (the start block varies by table) plus real time from the same endpoint, so one query can run from history to the chain head without a separate streaming pipeline. Devnet data is available from block 360,000,000.

That makes it practical to pull, say, every SPL Token transfer touching one program across years of history with plain HTTP requests, which is slow and costly through standard RPC methods. Solana is also SQD's busiest chain: the company reported 2.8 billion data requests across the network in the twelve weeks to September 11, 2026, of which 1.12 billion were for Solana.

Squid SDK and Pipes SDK: building a Solana indexer

SQD ships two open-source TypeScript SDKs that read from Portal:

• Squid SDK is the established full-stack framework: a batch processor, TypeORM persistence, a GraphQL API and a CLI to deploy to SQD Cloud. For Solana it supports instruction, transaction, balance and reward streams, with a typegen tool for type-safe instruction decoding.
• Pipes SDK is the lighter, composable option: a chainable query builder that streams data into Postgres, ClickHouse, MongoDB, SQLite, BigQuery, Parquet or a custom target, handling pagination, decoding, reorgs and resume logic. Pipes SDK 1.0 is on npm under a beta tag, covering EVM, Solana, Bitcoin, Tron and Hyperliquid fills.

Neither requires running a validator, an archive node or your own RPC. SQD also publishes an MCP server and agent skills so AI tools such as Claude, Cursor and VS Code can query onchain data directly, which is the basis of its "data layer for AI agents" positioning.

SQD Portal pricing and the free public endpoint

According to SQD's pricing docs, Portal has a free rate-limited public tier (roughly 50 MB a month across all chains), a free Testing tier with an API key, Starter at $99 a month, Growth at $499 a month and custom Enterprise plans. Paid plans are a flat monthly price with no compute units, per-request fees or egress charges, and are priced in US dollars. As of October 2026 the docs say Portal billing "is not live yet": paid plans run through early access while the public endpoint keeps its existing limits.

SQD Cloud, SQD 360 and the Google Cloud BigQuery partnership

• SQD Cloud is the managed platform for deploying and scaling squids (indexers) without running servers.
• SQD 360 is the enterprise arm that builds custom data pipelines and dedicated Portal instances.

On August 24, 2026 SQD announced a partnership with Google Cloud Web3: through SQD 360 it supplies the indexing and pipelines behind Google Cloud's blockchain datasets in BigQuery. The launch covered 10 chains from genesis, with more chains and "agentic" features planned. SQD's announcement does not say whether Solana is among the ten.

Other named users include Morpho, GMX, PancakeSwap, RAILGUN, zkVerify and Deutsche Telekom. Turtle runs its cross-chain opportunity index on Portal, and Olas backfilled roughly 14 million Polygon blocks in about a week in September 2026, after a subgraph rebuild had been estimated at two to three months.

SQD token: worker staking, delegation and Revenue Pools

SQD is an ERC-20 token. Its main contract is on Arbitrum (0x1337420dED5ADb9980CFc35f8f2B054ea86f8aB1), with deployments on Ethereum, Base and BNB Chain. There is no official SQD token on Solana, so treat any Solana "SQD" mint with suspicion. The token has four jobs:

• Worker bonds: running a worker requires bonding 100,000 SQD on Arbitrum, plus a machine with 4 vCPU, 8 GB RAM and a 1 TB SSD. The bond is locked for about two weeks of Ethereum blocks and can be slashed for provable misbehaviour.
• Delegation: holders can delegate to workers they trust and share their rewards. The whitepaper targets about 10% a year for delegators and 20-30% for workers, depending on bonded capital; these are design targets, not guarantees.
• Bandwidth: gateway operators lock SQD to get higher query rate limits, with longer locks earning a multiplier.
• Governance participation.

Revenue Pools, launched in limited beta on December 31, 2025, link the token to paying customers. Enterprise clients pay in stablecoins and SQD holders who lock tokens can share in those payments. In the public pool's design, a FeeRouter splits fees 50/45/5: half stays as stablecoins for pool providers, 45% is swapped to SQD on PancakeSwap V3 (with a TWAP oracle guarding against manipulation) for worker rewards, and 5% is swapped and burned. SQD stresses that pool payouts are not an entitlement and can be changed or suspended at any time.

Rezolve Ai has also said it intends to buy SQD worth at least 1% of its annual revenue, and that the token will be rebranded once regulators and exchanges approve. No rebrand had happened as of October 2026.

Who SQD Network is for

• Developers building Solana analytics, dashboards or backends that need deep history and decoded tables rather than raw RPC calls. The free public Portal is enough to prototype; production needs a paid plan or dedicated Portal.
• Multi-chain teams that want Solana and EVM data through one API and one SDK, instead of a separate vendor per chain.
• Enterprises and data teams that need a validated, auditable feed, through SQD 360 or Google BigQuery.
• AI agent builders who want onchain data available through MCP.

It is a poor fit for latency-critical Solana trading bots, which need raw Geyser streams, and for anything that has to send transactions: SQD is read-only.

Is SQD Network safe? Risks and limitations

• Centralised in practice, for now. Workers are permissionless, but Subsquid Labs is still the primary data provider and runs the scheduler that assigns data to workers. Errors at ingestion would propagate before any worker-level check applies.
• Snoopy is beta. The fraud-proof and slashing system is SQD's main guarantee against bad worker data, and it is not yet final.
• Corporate ownership. Since October 2025 the company behind the network is a subsidiary of Rezolve Ai (Nasdaq: RZLV), so roadmap and token decisions, including the planned rebrand, follow a listed parent's strategy.
• Token and contract risk. Worker bonds, delegation and rewards run on Arbitrum smart contracts, not Solana. The contracts are open source, but we did not find a published third-party audit of them or a statement of who can upgrade them in SQD's docs. Delegators also carry SQD price risk and slashing risk on the worker they back.
• Pricing not yet live. Paid Portal plans are in early access, so production costs and limits may still change.
• Self-reported metrics. The benchmarks, request counts and client figures above come from SQD or its customers.

SQD vs The Graph, Triton One and other alternatives

SQD vs The Graph: the most-searched comparison. SQD's own comparison frames SQD as a raw-data lake read through TypeScript SDKs, where The Graph serves indexed data via subgraphs written in AssemblyScript and paid for in GRT on its decentralised network. SQD covers far more chains and offers a fiat-priced tier. The Graph has the larger library of existing subgraphs and a longer-running decentralised network. The Graph's Solana support comes through its newer Token API rather than classic subgraphs.

SQD vs Triton One and Helius: these are Solana RPC and streaming providers, not direct substitutes. SQD concedes that Triton One is the better choice for transaction sending, sub-slot Geyser gRPC streams and its Old Faithful Solana archive. SQD's edge is decoded, cross-chain history with no egress fees. Many teams run an RPC provider for live reads and writes and SQD for indexing and backfills.

Other alternatives: Dune for SQL analytics on decoded tables, plus data APIs such as Goldsky, Bitquery, Allium, Alchemy and QuickNode. Pick Dune for ad-hoc analysis and SQD when you are building an indexer or pipeline that feeds your own database.

Rezolve Ai acquisition, team and history

Subsquid was founded in 2021 by Marcel Fohrmann and Dr. Dimitrii Zhelezov (now CTO), and rebranded to SQD in July 2024. Rezolve Ai announced the acquisition on October 9, 2025 and closed it on October 10. Rezolve said the Subsquid and Smartpay deals together cost about $10 million in cash plus 1 million Rezolve shares, with SQD as the data layer of its "agentic commerce" stack. SQD said at the time that the team and the network's fundamentals would stay unchanged.

In a June 24, 2026 repositioning towards institutional clients, investor Wanja S. Oberhof became CEO and former EY partner Robert Jung joined as Head of Transformation, with Fohrmann continuing to lead protocol and ecosystem work. The board is chaired by Davey Airey and includes Dr. Marcel Reichart of Rezolve Ai, Jordi Alexander of Selini, Sebastian Weiss, Jesus Martinez and Clemens Jakopitsch. At that point SQD said it served up to 1.4 petabytes of data a day, up from 240 terabytes in February 2026, across 225+ indexed networks.