The code does not lie, but it does hide. On stage at Devcon VI, Vitalik Buterin dropped a bomb that most attendees missed amid the buzz of zk-rollups and restaking. He said: "Every time you query a node through Infura or Alchemy, you're handing over metadata about your application, your users, and your strategy. And that data is being used to train better products for them, not for you."
The hook is subtle but devastating. It’s not about smart contract bugs. It’s about the invisible data tax that blockchain applications pay to the infrastructure layer every single day. Vitalik didn’t name names, but the target is clear: the centralized RPC providers that power 80% of Ethereum dApps.
Let’s rewind. Blockchain promised self-sovereignty—ownership of your assets, your identity, your data. But in practice, most users and developers outsource the critical gateway—the RPC endpoint—to Infura, Alchemy, QuickNode, or similar. The ledger is public, sure. But the query patterns, the labels, the time series of interactions—that’s proprietary. And it flows straight into the infrastructure provider’s data lake.
Context: The Stargate Paradox
The parallel to Satya Nadella’s recent warning about AI model providers is uncanny. Just as Microsoft CEO argued that enterprises pay token costs while losing their proprietary knowledge to model training, the same dynamic exists in blockchain. Developers pay gas fees and subscription costs for RPC services, but the real value they generate—the behavioral fingerprint of their entire application—becomes an asset of the infrastructure provider. It’s a stargate: you pay to enter, but your map is sold behind your back.
Alpha hides in the friction of liquidity. The liquidity here is data flow. The friction is the lack of ownership over the query trail. Vitalik’s call to action is simple: “Stop renting your data. Start owning your infrastructure.” But the industry is addicted to convenience.
Core: The Order Flow of Information
Let’s get technical. Every Ethereum transaction starts with a query—a user checks a balance, a dApp loads state, a bot simulates a trade. That query hits an RPC endpoint. The provider logs: IP, request payload, timestamp, response time, error codes. Over months, these logs form a unique signature of the application’s lifecycle. An exchange can infer trading volume patterns. A lending protocol reveals liquidation triggers. An NFT marketplace exposes hot wallet activity.
From my experience auditing DeFi applications in 2022, I saw a project migrate from public Infura to self-hosted Geth nodes because they discovered that their internal arbitrage bot’s queries were being routed through Alchemy’s enhanced API. They paid for the premium tier, yet they had no visibility into how Alchemy used their query data to optimize its own MEV protection features. The code does not lie, but it does hide.
Vitalik proposes a three-layer separation: the execution client, the data availability layer, and the RPC middleware. By decoupling the query path from the trusted provider, applications can retain control. Use a self-hosted archive node for critical state, a decentralized RPC network like Pocket or Blast for fallback, and a local caching layer for frequent reads. This is the blockchain equivalent of what Nadella calls “owning your evaluation, memory, and operation logs.”
Contrarian: The Retail vs Smart Money Data Gap
Retail dApps cheer for low-latency RPCs and cheap API keys. They don’t read the terms of service. Smart money—large protocols, quant funds, and institutional custodians—already build their own query infrastructure. They run their own Erigon nodes, subscribe to direct data feeds from validators, and encrypt their RPC traffic with TLS+zero-knowledge proofs. The gap is widening. Retail dApps are unknowingly feeding the very beasts that will later compete with them via superior analytics.
Backtest the assumption, not just the data. The assumption is that infrastructure providers are neutral pipes. They aren’t. Every major RPC provider now offers “analytics dashboards” and “user behavior insights” as premium products. Guess where the data for those products comes from? Your live traffic. You are the product waiting to be mined.
Vitalik’s solution—self-hosted nodes for critical queries—isn’t new. What’s new is the urgency. With the rise of account abstraction (ERC-4337) and intent-based architectures, the RPC layer is becoming the execution environment itself. If you outsource that, you outsource the brain of your application.
Takeaway: The Data Sovereignty Premium
In the next bull run, attention will shift from TVL and yield to data sovereignty. Protocols that prove they don’t leak query patterns will command a premium. Expect a new rank: “Node self-reliance score.” The exchange that runs its own full nodes will trade at a higher multiple than one that plugs into Infura.
Volatility is the tax on uncertainty. Right now, the uncertainty is whether your infrastructure provider will turn your data into their competitive edge. The only hedge is to own the pipeline—from the genesis block to your user’s last click. Start running your own node. Not tomorrow. Today.