RPC Providers Compared
How RPC providers like Alchemy, Infura, QuickNode, and Ankr sit between wallets and blockchains — their business models, reliability differences, censorship risks, and how to think about decentralised alternatives.
The basics
What an RPC Provider Actually Does
Every time you click 'Swap' in a wallet, your wallet has to ask a blockchain node what your balance is, what the current gas price is, and finally submit the signed transaction. That conversation is called an RPC (Remote Procedure Call) request. Wallets, dApps, and block explorers all rely on RPC endpoints to talk to blockchains. Running your own node is expensive and technically demanding, so almost everyone uses a hosted RPC provider — Alchemy, Infura (owned by ConsenSys), QuickNode, Ankr, Chainstack, or a smaller specialist. When you install MetaMask, its default Ethereum RPC is Infura; when you use Uniswap in a browser, the requests flow through one of these providers.
The Free Tier and the Reality
Most providers offer generous free tiers so that developers and hobbyist users never see the bill. The revenue comes from dApps and wallets that hit rate limits, need advanced features (archive nodes, trace APIs, subscription websockets), or want SLAs. This creates a two-tier reality: retail users interact with a small number of providers indirectly, and those providers are load-bearing for the whole ecosystem. A single Infura outage on 11 November 2020, which Infura itself called its most severe service interruption in four years of operation, disrupted many of the products built on it.
Why This Matters for Decentralisation
Blockchains are decentralised at the protocol level, but access to them is often centralised at the RPC layer. If a handful of RPC providers went offline simultaneously, most wallets would appear broken even though the chains themselves would keep producing blocks. Providers can also, at least in theory, filter transactions, log user activity, or geo-block users. This risk has driven interest in decentralised RPC networks (Pocket Network, Ankr's decentralised endpoints) and in wallets that let users configure their own RPC endpoints.
- RPC providers are the translation layer between wallets/dApps and blockchains
- Most retail users never choose their RPC provider explicitly — the wallet picks
- Free tiers cover retail load; paying customers subsidise the infrastructure
- RPC centralisation is a real practical risk even when the blockchain is decentralised
Key Takeaways
- Every wallet interaction with a blockchain goes through an RPC endpoint
- A small number of providers (Alchemy, Infura, QuickNode) sit behind many wallets and dApps
- Outages at these providers can make the whole ecosystem look broken
- You can usually override the default RPC in your wallet's advanced settings
Going deeper
How Providers Actually Differ
On the surface, RPC providers look interchangeable — they all speak the same JSON-RPC vocabulary. In practice they differ on a few axes. Reliability: published uptime SLAs differ by plan (several providers advertise 99.99%); multi-region redundancy and failover quality vary. Feature depth: archive nodes (needed to query historical state), debug traces (needed for transaction simulation), and specialised endpoints (e.g., Flashbots Protect RPC to avoid MEV) are premium features. Chain coverage: all majors cover Ethereum + top L2s; broader multi-chain coverage varies widely. Pricing model: some price per request, some per compute unit, some per bandwidth — same load can cost very different amounts depending on how usage is measured.
Wallet-Provider Relationships
MetaMask ships with Infura as the default Ethereum endpoint because both are ConsenSys projects. Other wallets make different choices. When a wallet routes millions of daily requests through a specific provider, that relationship is commercially important on both sides and creates lock-in that's hard to see from the outside. This is why a provider outage tends to affect specific wallets rather than the whole ecosystem uniformly.
Custom RPC — When and Why
Advanced users can point their wallet at a custom RPC endpoint for practical reasons: to avoid a specific provider (privacy or trust concerns), to route through Flashbots Protect (avoiding sandwich attacks), or to use a specific L2 endpoint that offers lower latency. This does not require running your own node — you can grab endpoints from providers, from public health-check aggregators like chainlist.org, or from the L2 team's own documentation. The default endpoint is a convenience, not a requirement.
Public Endpoints Are Not Free Infrastructure
Public 'free' RPC endpoints (the ones aggregated on chainlist.org) are often subsidised by providers hoping to convert users to paid plans, or by the chain's foundation. They are best-effort and can be rate-limited or shut off. If you're building anything that depends on RPC availability, use a paid plan or run your own node — never rely on public endpoints for production traffic.
Key Takeaways
- Providers differ on reliability, feature depth, chain coverage, and pricing model
- Wallet defaults create commercial relationships that affect who breaks together
- You can override the default RPC in most wallets — useful for MEV protection or trust reasons
- Public endpoints are best-effort — do not build production apps on them
Under the hood
Archive Nodes and Their Cost Structure
A full node keeps recent state; an archive node keeps every historical state at every block. Archive data for Ethereum is now multiple terabytes and grows constantly. Running an archive node yourself requires expensive NVMe storage, high-bandwidth ingest, and continuous re-syncing after upgrades. Providers charge substantial premiums for archive access because the underlying hardware is expensive. Applications that need historical queries (indexers, analytics platforms, block explorers, some governance tools) usually rely on providers rather than run their own — and those applications bear the direct economic cost of RPC pricing.
Decentralised RPC Networks
Pocket Network (POKT), Ankr's decentralised endpoints, and similar projects aim to distribute RPC serving across many independent nodes coordinated by token incentives. The pitch: reduce single points of failure, remove any single provider's ability to censor. The reality is more nuanced — quality of service depends on the network's node operators, and coordinating tens of thousands of nodes to serve millions of requests at sub-100ms latency is genuinely hard. MetaMask, for example, still defaults to a centralised provider (Infura).
MEV-Aware RPCs
Flashbots Protect and similar MEV-aware RPC endpoints route your transactions to private orderflow rather than the public mempool. This prevents sandwich attacks and front-running against your swaps. The trade-off: transactions can take slightly longer to include, and you're trusting the private-orderflow provider not to use your transactions for its own advantage. For DEX traders moving meaningful size, MEV-aware RPCs are generally worth using — they keep transactions out of the public mempool, away from front-running and sandwich bots.
The Long-Term Architecture Question
Where RPC serving lands over the next 3-5 years is genuinely uncertain. Options include: continued centralisation (a few providers get bigger, better, and more efficient); serious decentralised RPC networks (POKT-style tokens make economic sense at scale); protocol-level improvements (built-in light-client-friendly primitives reduce reliance on trusted providers); or vertical integration (major wallets and chains run their own RPC infra). Which mix wins matters for censorship resistance, uptime, and the geographical distribution of blockchain access.
How to Audit Your Own RPC Exposure
For any dApp or wallet you rely on, ask: which RPC provider does it use by default? What happens if that provider goes down? Do I have a fallback configured? If you're deploying something to production, use at least two providers with automatic failover. If you're a user of a wallet you rely on for meaningful value, know how to switch RPCs in an emergency — the two-minute answer to a provider outage is almost always to switch to a different endpoint, not to wait for the outage to resolve.
Key Takeaways
- Archive nodes are expensive to run; providers charge premiums accordingly
- Decentralised RPC networks exist but haven't yet displaced centralised providers at scale
- MEV-aware RPCs are worth using for DEX traders moving real size
- Long-term RPC architecture is genuinely uncertain — worth watching
- Always know how to switch RPCs; every wallet supports it in some form
Related Content
Related Coins
Key Terms
More Topics
Browse all 189 topicsValidium, Plasma, and State Channels
The scaling designs that lost to rollups — Validium (off-chain data), Plasma (mass exit games), and state channels (bilateral off-chain) — and why understanding them helps make sense of current rollup design.
App-Chains and Rollup-as-a-Service
How Cosmos-style app-chains and modern Rollup-as-a-Service (RaaS) providers (Caldera, Conduit, AltLayer, Gelato) let applications deploy dedicated chains, and the trade-offs vs shared chains.
Cosmos IBC and Interchain Security
How Cosmos IBC enables cross-chain messaging without wrapped assets, how Interchain Security (ICS) let chains rent security from the Cosmos Hub before the Hub removed it in 2026, and what these primitives mean for Cosmos's app-chain thesis.
Polkadot Parachains
How Polkadot's parachain model works — shared security via the Relay Chain, XCM cross-chain messaging, agile coretime — and how the network has evolved from parachain auctions to coretime markets.
Indexers and The Graph
How blockchain indexers turn raw chain data into fast, queryable APIs — The Graph's subgraphs, competing indexers, why every dApp uses one, and how the indexing layer shapes what applications can build.
Oracle Providers Compared
Chainlink, Pyth, RedStone, API3, and Tellor — how each oracle network sources and delivers off-chain data, their trust and economic models, and which use cases suit which oracle.
Bridge Architectures Compared
How Wormhole, LayerZero, Hyperlane, Axelar, and Across handle cross-chain messaging and asset transfer — their trust models, exploit histories, and the fundamental trade-offs between speed, security, and generality.
CBDCs by Jurisdiction
The state of central bank digital currency projects across China, EU, UK, US, and other major jurisdictions — retail vs wholesale designs, privacy trade-offs, and how CBDCs interact with existing stablecoins.
References & further reading
- primary
- secondary
- primaryMetaMask Help Center - What is Infura, and why does MetaMask use it?
MetaMask's default RPC provider and how to switch.
- primaryFlashbots Docs - Protect quick start
How MEV-aware private-orderflow RPCs work.