We decided to move away from the Cosmos tech stack for a few reasons.
First, we have a lot of respect for Tendermint/CometBFT and are grateful for how well it has powered Noble to date. That said, consensus research has moved forward significantly, and adopting Simplex (via Commonware) lets us stay at the forefront of that innovation while preserving the security and finality guarantees we’re accustomed to. We’re deliberately optimizing for next-generation consensus research and implementations, and Simplex (2023) aligns better with where we want to be long-term than Tendermint (2010) like algorithms.
Simplex preserves instant finality under the same ≥2/3 honest-stake assumption as Tendermint, but achieves it with a simpler, more efficient consensus core. It reduces round complexity, avoids leader churn and timeouts in the common case, and is designed for higher throughput and lower latency, especially under stable network conditions.
Second, while Cosmos EVM is a meaningful step in the right direction, integrating the EVM directly via a battle-tested execution client like Reth gives us substantially more flexibility and control over the execution layer than is practical within the Cosmos SDK today.
Hope that makes sense.