BM's New EOSIO Proposal: Can It Solve the CPU Crisis and Save EOS?

The EOS Congestion Nightmare
For over a month, EOS users have faced crippling network congestion, with DAppTotal data showing 77.76% of CPU resources consumed by the token-mining project EIDOS. The situation became so dire that EarnBetCasino threatened to abandon the chain entirely—a stark reminder that even ‘high-performance’ blockchains aren’t immune to resource wars.
BM’s Diagnosis: A Design Flaw
In his recent proposal, EOS creator Daniel Larimer (BM) identifies the root issue: REX (Resource Exchange), the system designed to rent out CPU resources, operates on flawed assumptions. It expected normally distributed demand but got hit by Pareto distributions instead—where a few actors dominate usage. The result? Prices swing wildly, and at times, no CPU is available at any price.
The Band-Aid That Failed
Previous fixes like “partial reserve CPU” provided temporary relief but introduced new problems:
- 1000x CPU boosts during low usage created false expectations
- Sudden demand spikes locked users out unpredictably
- Speculative pricing detached from actual utility value
The New Algorithm: Predictability Through Exponential Pricing
BM’s solution flips the script by:
- Making all CPU time lease-only (no permanent ownership)
- Implementing exponential price curves based on utilization
- Distributing rental fees to stakers (REX participants)
The math ensures stable pricing: if you want 1% of network CPU for 30 days, you pay the difference between current and projected rental income at that utilization level. Charts in his post show prices remaining reasonable until ~10% network usage.
Migration Challenges Ahead
Transitioning won’t be painless:
- Current stakeholders must adapt to losing “permanent” CPU rights
- REX markets will gradually phase out over ~12 months
- Early adopters may face temporary disadvantages
As someone who’s analyzed dozens of DeFi fails, I’ll believe this works when I see on-chain adoption metrics—but credit to BM for acknowledging past miscalculations. If implemented well, this could make EOS competitive again… assuming developers haven’t already jumped ship to Solana.