RISC Zero

RISC Zero, founded in 2021 in Seattle by Brian Retford and team, has the cleanest execution record in the verification cohort, shipping the first general-purpose RISC-V zkVM, the Bonsai hosted prover, the Steel Ethereum-state library, and the Boundless mainnet in sequence. That is a multi-product roadmap delivered, not a single primitive, and it reflects serious engineering depth. Against EZKL's single library or Giza's pivot, RISC Zero has executed a coherent, expanding stack. The view would weaken on post-token organizational churn, and strengthens as Boundless accrues real prover and consumer activity.

The core differentiation is generality: a zkVM emulating the RISC-V instruction set lets developers prove arbitrary Rust code, so verifiable compute is not limited to hand-built circuits for one task. That breadth is exactly what makes it a threat to purpose-built zkML tools — arbitrary-code proving can, in principle, subsume bespoke ML circuits like EZKL's. Versus Succinct's SP1, also a strong RISC-V zkVM, the contest is close and turns on performance, tooling and ecosystem rather than concept. The view deepens on benchmarks showing competitive proving cost and latency at scale, and softens if a rival zkVM clearly out-performs.

ZKC launched at the Boundless mainnet in September 2025 with a one-billion supply and a Proof of Verifiable Work mechanism intended to pay provers for useful proving, a thoughtful design that ties emissions to real work in theory. In practice the token met sharp sell pressure around its Binance debut, with reports of a steep post-listing drop amid airdrop unlocking, so value accrual is unproven. Against pre-token peers like EZKL, RISC Zero now carries live-token obligations and price exposure. Lookout would turn positive on evidence that PoVW rewards track genuinely utilized proving demand, and negative if emissions outrun paid usage.

RISC Zero has the most production-relevant footprint in the cohort: Bonsai has served as a hosted prover for real integrations, the zkVM is widely referenced, and Boundless launched with prover participation and exchange listings. The open question is demand-side depth — whether the proof marketplace attracts sustained, paid proving jobs rather than incentive-driven supply. Against EZKL's experimental usage or Giza's agent volume, RISC Zero's traction is broader but its marketplace is freshly live and unproven on durable demand. The axis moves up on utilized GPU-hours and paying proof consumers, and stays neutral while activity is early.

RISC Zero has raised roughly $52M, anchored by a $12M seed led by Bain Capital Crypto in 2022 and a $40M Series A led by Blockchain Capital in 2023, with Galaxy, Fenbushi, Delphi and IOSG among participants. That is among the best-capitalized cap tables in the verification sector and signals tier-one conviction in the verifiable-compute thesis. It dwarfs EZKL's seed and exceeds Giza's total, giving RISC Zero runway and credibility through a long infrastructure build. The view would weaken only on a down-round or post-token investor fatigue; for now funding is a clear strength.

RISC Zero owns the 'universal zero-knowledge' framing — verifiable compute for any program — which is a larger, more durable narrative than zkML specifically, and it captured significant mindshare as the zkVM race became a defining 2024-25 infrastructure theme. Backing from the Ethereum Foundation and a Base mainnet for Boundless reinforce its position at the center of the proving narrative. Against EZKL's niche zkML story or Giza's crowded DeFAI pivot, RISC Zero's framing is broad and investable. The narrative stays strong while verifiable compute is treated as foundational, and would soften only if the zkVM thesis lost momentum to alternative scaling approaches.

The principal risks are token-and-demand, not credibility: ZKC's heavy post-launch sell pressure raises questions about incentive sustainability, and a proof marketplace must convert subsidized capacity into genuine paid demand before emissions normalize. The category is also crowded and capital-heavy, with Succinct and a field of coprocessors competing for the same developers. Relative to dormant Modulus or tool-only EZKL, RISC Zero's risk is market-structural rather than existential — it has the product and the backing, and must now prove demand. Lookout would de-risk on real proving revenue and stable token economics, and flag it upward if Boundless usage stays incentive-dependent.