KoinlyticsVitalik grouped every future Ethereum change into five phases, each a play on words. Two of them have shipped enough to already matter. The other three are the reason the chain is still evolving.
The Merge (2022)
Ethereum switched from proof of work to proof of stake in September 2022. Overnight, energy use dropped by ~99.95%. Issuance fell from around 5 million ETH per year to under 1 million. Combined with EIP-1559 fee burning, the net supply of ETH now shrinks in most months.
The Surge (ongoing)
Scaling to 100,000+ TPS via rollups. Dencun in March 2024 added blob transactions (EIP-4844), which cut Layer 2 costs by 10-100x. Pectra, Fusaka, and further blob capacity increases are the next legs. Danksharding is the endgame: sharding data availability across the whole validator set.
The Verge (in design)
Making a full node runnable from a phone. Verkle trees replace Merkle trees for state, letting nodes verify without holding the entire state. Stateless clients are the goal. If this ships, running a node stops requiring a beefy machine, which strengthens decentralization.
The Purge (in design)
Deleting historical state that clients no longer need to hold. Ethereum currently forces every full node to store every transaction since 2015. That is unsustainable. Purge introduces history expiry: after a year, old data lives on portable networks and archive nodes, not on your laptop.
The Splurge (grab bag)
Everything else. Account abstraction (ERC-4337 already live), single-slot finality, EVM improvements, MEV-resistance research, quantum readiness. Roadmap treats this bucket as necessary but non-blocking.
What the market cares about
Rollups. That's it, honestly. Every other item on the roadmap is a means to make rollups cheaper, safer, or more decentralized. If you understand why Ethereum bet on rollups instead of monolithic sharding, you understand the roadmap.