Hoskinson might be wrong about the future of decentralized compute

At a recent Consensus event, Charles Hoskinson, founder of Cardano, addressed concerns about the blockchain trilemma and the role of hyperscalers like Google Cloud and Microsoft Azure in decentralization. He argued that major blockchain projects require these large cloud providers and that risks can be mitigated through advanced cryptography, multi-party computation (MPC), and confidential computing, asserting that if the cloud cannot see the data, it cannot control the system.

Limitations of MPC and Confidential Computing

While technologies like MPC and confidential computing offer significant security enhancements, they do not entirely eliminate underlying risks. MPC distributes key material across multiple parties, reducing the risk of a single compromised node but expanding the security surface to include coordination, communication channels, and governance. Confidential computing, through trusted execution environments (TEEs), encrypts data during execution, limiting exposure to hosting providers. However, TEEs rely on hardware assumptions and have demonstrated vulnerabilities to side-channel and architectural attacks. Crucially, both MPC and TEEs often operate on top of hyperscaler infrastructure, leaving the physical hardware, virtualization layer, and supply chain concentrated. This concentration grants infrastructure providers operational leverage, allowing for potential throughput restrictions, shutdowns, or policy interventions, even if data inspection is prevented by cryptography.

The 'No L1 Can Handle Global Compute' Argument Re-examined

Hoskinson's assertion that hyperscalers are necessary due to the computational limitations of Layer 1 (L1) networks is partially correct regarding L1's primary function: maintaining consensus and state transitions, not running global-scale computations like AI training. However, the core issue for global systems is the verifiability of results, regardless of where computation occurs. Modern crypto infrastructure increasingly relies on off-chain computation, with the critical element being the ability to prove and verify results on-chain, as seen in rollups and zero-knowledge systems. The true concern lies not in L1's computational capacity but in who controls the execution and storage infrastructure supporting verification. Dependence on centralized infrastructure for off-chain computation introduces centralized failure modes, even if settlement remains decentralized in theory.

Infrastructure Dependency vs. Cryptographic Neutrality

Cryptographic neutrality, the idea that protocol rules cannot be arbitrarily changed and backdoors cannot be introduced, is a powerful concept. However, cryptography relies on hardware, and the physical layer dictates participation accessibility and cost. If hardware production, distribution, and hosting remain centralized, participation becomes economically gated, undermining the protocol's theoretical neutrality. In high-compute systems, hardware is paramount, influencing cost, scalability, and resilience. A neutral protocol running on concentrated infrastructure is practically constrained. The priority should shift towards combining cryptography with diversified hardware ownership to ensure true neutrality and fair participation, as concentrated infrastructure creates fragility and allows providers to impose leverage through rate-limiting or policy interventions.

Specialization in Compute Markets

Competing with hyperscalers like AWS is often misconstrued as a battle of scale. Hyperscalers optimize for flexibility and general-purpose compute, which involves significant cost layers. Specialized systems, such as those for zero-knowledge proving or verifiable compute, are designed for specific, compute-dense tasks. These systems, when vertically integrated across hardware, software, and design, achieve greater efficiency in terms of proof per dollar, watt, and latency. Purpose-built networks outperform general-purpose cloud providers for steady, high-volume tasks by optimizing for structural efficiency and sustained utilization rather than short-term rental models. This specialization leads to a different economic structure, better aligned with protocol incentives.

Strategic Use of Hyperscalers

Hyperscalers are valuable infrastructure providers, but dependence on them poses a risk to decentralization. A resilient architecture should utilize hyperscalers for burst capacity, geographic redundancy, and edge distribution, rather than anchoring core functions to them. Critical functions like settlement, final verification, and the availability of essential artifacts should remain operational even if a cloud region fails or a vendor exits a market. Decentralized storage and compute infrastructure offer a viable alternative for housing proof artifacts and verification inputs, ensuring they are not subject to a provider's discretion. Hyperscalers should serve as optional accelerators, providing reach and burst capabilities, but the system's core proof generation and persistence capabilities should not be gated by a single vendor. This approach fortifies crypto's ethos of decentralization by ensuring the network can continue functioning even if a major cloud provider disappears.