Solana simulation evaluates SGP-03 fee impact on major apps and routers

Solana’s fee model is about to get a lot more granular, and not every application will love the math. A simulation run by analyst @MostlyData_ quantifies what SGP-0003 would actually do to transaction costs across the network, and the results paint a picture of winners, losers, and a dramatically larger burn rate. The governance proposal, introduced on August 3, 2026, alongside SIMD-0553, would retire Solana’s flat 5,000-lamport base fee in favor of a two-part system: a fixed 2,500-lamport inclusion fee paid to the block leader, plus a variable resource fee starting at 0.1 lamports per requested compute unit that gets burned entirely. What the simulation actually found The core finding is blunt: Solana transactions are ordering way more food than they eat. On average, transactions request about 20% more compute units than they actually consume. Under the current flat-fee system, that overestimation costs nothing extra. Under SGP-0003, it costs real money. That mismatch hits routers and aggregators hardest. The simulation pegged average fee increases at roughly 0.000068 SOL for Jupiter, 0.00010 SOL for Titan, and 0.00012 SOL for DFlow. Only about 28% of transactions would see a fee ...