a16z Upgrades Jolt zkVM With Genuine Privacy Layer, Takes Aim at Industry's ZK Labeling Habits
a16z Crypto on March 3 updated its open-source Jolt virtual machine with a new cryptographic component that delivers genuine privacy protections, while publicly arguing that most of the industry has applied the "ZK" label to systems that do not actually deliver zero-knowledge privacy.
The venture firm integrated a technique called NovaBlindFold into Jolt, a zero-knowledge virtual machine (zkVM) it has developed since 2023. A zkVM is a system that can execute a program and produce a compact cryptographic proof that the computation ran correctly, without requiring anyone to re-run the computation themselves. The NovaBlindFold addition gives Jolt a property that its name implies but, according to a16z, most similar systems lack: the prover cannot leak any information about the private inputs used during the computation.
The Terminology Problem
"Most zkVMs are not actually zero knowledge, unless an expensive 'wrapping' procedure is applied," a16z stated in conjunction with the update.
The firm draws a clear line between two properties that the industry routinely conflates. The first is succinctness: proofs are short and fast to verify. The second is zero-knowledge: the proof reveals nothing about the underlying data. Almost all zkVMs currently deliver the first property. Very few deliver the second without bolting on additional steps.
The "wrapping" workaround a16z describes involves recursively proving the verification of one proof system inside a second, genuinely private proof system. This approach is computationally expensive and typically requires a trusted setup, a one-time ceremony that introduces trust assumptions into the system. NovaBlindFold sidesteps this entirely. The privacy overhead it adds to Jolt is approximately 3 kilobytes above the base proof size.
The criticism lands in a competitive market. Rivals including RISC Zero and Succinct's SP1 have faced similar questions about whether "ZK" is accurate shorthand or a marketing choice. The zkVM sector collectively represents roughly $11.7 billion in project market capitalization, according to a Gate.io ecosystem report. A separate BingX/AInvest ZK report puts total value locked across ZK-based rollups at more than $28 billion.
What Jolt Is and Where It Came From
Jolt was first described in the academic paper "Jolt: SNARKs for Virtual Machines via Lookups," published in August 2023 and co-authored by Arasu Arun of NYU, Srinath Setty of Microsoft Research, and Justin Thaler, a researcher at a16z and professor at Georgetown University. The paper was presented at Eurocrypt 2024. The open-source codebase, written in Rust and running on the RISC-V instruction set, was released in April 2024. At under 25,000 lines of Rust, the codebase is compact enough to make thorough security auditing practical, a meaningful credibility signal in this space.
Thaler described the purpose of zkVMs when Jolt launched: "[zkVMs] scale blockchains by kind of doing the hard work off-chain, and only having the blockchain verify the proofs."
The system's proof size has improved substantially since early development, dropping from several megabytes to around 200 kilobytes, with a target of 25 kilobytes on the roadmap alongside roughly a 3x improvement in prover speed.
NovaBlindFold is a variant of the Nova folding scheme, a technique originally developed at Microsoft Research. Folding schemes reduce two separate NP statements into one, enabling long computations to be proved incrementally without the proof growing in size at each step.
Regional Stakes: India and Africa
The update has practical implications well beyond North American developer circles.
In India, the national identity authority UIDAI outlined an Aadhaar Vision 2032 framework in late 2025 that explicitly references blockchain, quantum computing, and advanced encryption in the context of protecting data for over 1.4 billion residents.
The framework connects to India's Digital Personal Data Protection Act of 2023. Analysts argue that genuinely zero-knowledge proofs would be well-suited to the core challenge the system faces: confirming identity without exposing the underlying biometric data. The Aadhaar Vision 2032 framework does not explicitly name zero-knowledge proofs in its public-facing documents; the connection is an analytical inference based on the stated privacy-by-design goals of the system.
a16z's clarification that systems marketed as ZK were not delivering the privacy property by default is directly relevant to anyone evaluating cryptographic tools for national-scale identity infrastructure. Jolt's Rust-first design also aligns with India's growing systems programming developer community.
In sub-Saharan Africa, the stakes center on financial data. Nigeria received $19.5 billion in remittances in 2023 according to World Bank figures, and Zone, a blockchain payment network operating in the country, has processed over 1 trillion naira (approximately $636 million) in transactions.
Kenya led African startup funding in 2024 at $638 million. Researchers working on ZKP applications in Africa have identified privacy-preserving credit scoring, KYC verification via cryptographic proof, mobile money authentication, and DeFi smart contracts as leading use cases for zero-knowledge technology.
The minimal proof size overhead from NovaBlindFold is particularly relevant in markets where compute and bandwidth costs constrain small fintech operators. Two significant barriers remain: a shortage of developers with ZK expertise across the continent, and the unfamiliarity of many African regulatory bodies with ZKP-based compliance solutions, a gap that is especially consequential given the growing policy scrutiny around privacy-preserving infrastructure.
What Comes Next
a16z has stated it will not monetize the Jolt codebase. The firm's framing of the NovaBlindFold update as a correction to an industry-wide mislabeling habit signals that the terminology debate is likely to intensify as zkVMs become more embedded in blockchain infrastructure. For developers building privacy-sensitive applications, the update provides an open-source tool that makes the ZK label technically defensible without wrapping costs. The performance targets cited during Jolt's development (a 25-kilobyte proof size and roughly a 3x improvement in prover speed) remain on the roadmap, though a16z has not publicly stated a timeline for reaching those milestones.