Hoskinson might be wrong about the future of decentralized compute

The article challenges Charles Hoskinson's assertion that hyperscalers like Google Cloud and Microsoft Azure are not a risk to decentralization in blockchain compute, despite advanced cryptography like Multi-Party Computation (MPC) and Confidential Computing (TEEs). While MPC reduces single-node risk and TEEs encrypt data, they introduce new dependencies on coordination layers, hardware assumptions, and the underlying centralized infrastructure of hyperscalers. Cryptography prevents data inspection but not operational leverage like throughput restrictions or policy interventions.

The author contends that the focus should shift from Layer 1 computational capacity to who controls the off-chain execution infrastructure, as centralized compute leads to centralized failure modes, even if settlement is theoretically decentralized. Furthermore, while protocols can be cryptographically neutral, hardware concentration creates economic barriers to participation. Specialized, vertically integrated proving networks are argued to outperform generalized hyperscaler infrastructure for dense, deterministic workloads like zero-knowledge proving.

The conclusion advocates for using hyperscalers for burst capacity and redundancy, but not as the foundation for core functions like settlement and verification. True resilience requires decentralized storage and compute infrastructure economically aligned with the protocol, ensuring that critical artifacts are not subject to a single vendor's discretion or policy changes.

(Source：CoinDesk)