Skip to main content

ZKsync's Institutional Bet: How Five Regional Banks With $600B in Deposits Are Going On-Chain

· 10 min read
Dora Noda
Dora Noda
Software Engineer

Five U.S. regional banks holding over $600 billion in combined deposits are preparing to launch tokenized deposit accounts on a zero-knowledge Layer 2 blockchain — not as an experiment, but as a production payment network targeting customer availability by Q4 2026. The network is called Cari, and it runs on ZKsync's Prividium. It may be the clearest signal yet that ZKsync's pivot away from the consumer DeFi speed race and toward regulated financial infrastructure is paying off.

From "Fastest Cheap L2" to Institutional Infrastructure​

For most of its history, ZKsync competed on a familiar set of metrics: throughput, gas costs, EVM compatibility, and developer tooling. Those battles — fought against Arbitrum, Optimism, Base, and Polygon — produced an L2 ecosystem that now holds over $50 billion in total value locked across its leading chains.

ZKsync's share of that market is relatively modest. Arbitrum alone controls roughly 44% of L2 TVL at around $16.6 billion, while Base follows close behind.

In January 2026, Matter Labs published a roadmap that explicitly conceded the consumer DeFi speed race and announced a different game. The new strategy centers on a category ZKsync calls "real-world infrastructure" — regulated financial institutions, market infrastructure providers, and large enterprises that need blockchain's properties but cannot sacrifice confidentiality and compliance guarantees.

The pivot is structured around four commitments ZKsync describes as non-negotiable:

  • Privacy by default — transaction data hidden from public view by architecture, not configuration
  • Deterministic control — institutions retain governance over their chain parameters
  • Verifiable risk management — ZK proofs provide cryptographic audit trails
  • Native connectivity to global markets — Ethereum settlement without public data exposure

Each commitment is a direct rejection of what generic public L2s offer.

Prividium: The Privacy Rollup Built for Banks​

At the center of ZKsync's institutional strategy is Prividium — a private, permissioned Ethereum-secured Layer 2 that gives institutions the ability to execute transactions without exposing balances, counterparties, or internal decision-making logic to the public chain.

This is not a bolt-on privacy module. ZKsync baked confidentiality into Prividium's consensus layer, meaning that the architecture is structurally compliant with institutional data requirements from the ground up. The contrast with "privacy optional" approaches matters in regulated finance: a bank's legal team needs to certify that sensitive transaction data is never publicly visible by design, not by configuration.

Prividium still settles to Ethereum, preserving the security guarantees of the base layer without exposing internal activity on the public chain. Selective-disclosure audit channels let institutions share verified transaction proofs with regulators on demand, without broadcasting the underlying data to the world. That combination — privacy by default, verifiability on request — is precisely the architecture that MiCA's privacy-for-institutional-DeFi requirements demand and that optimistic rollup competitors cannot replicate at the architecture level.

The framework has been validated through live demonstrations with more than 35 financial institutions — including Citi, Mastercard, Deutsche Bank, and two undisclosed central banks — across cross-border payment corridors and intraday repo use cases.

The Cari Network: A $600 Billion Case Study​

The most concrete proof of the Prividium thesis landed in March 2026 when the Cari Network announced its design partner consortium. Founded by Eugene Ludwig — the 27th U.S. Comptroller of the Currency — Cari is building a bank-governed tokenized deposit platform on ZKsync's Prividium, with five named US regional banks participating: Huntington Bancshares, First Horizon, M&T Bank, KeyCorp, and Old National Bancorp. Their combined deposits exceed $600 billion.

The product is straightforward: tokenized deposits that move with the speed and programmability of stablecoins while remaining liabilities of the issuing bank. Unlike USDC or USDT, Cari tokens are not issued by a separate entity — they are bank liabilities in token form, which means they retain FDIC insurance eligibility and slot directly into existing bank compliance frameworks without requiring new regulatory approvals for the deposit product itself.

The competitive target is explicit. Headlines from CoinDesk, Bankless, and CryptoNews all called the launch a declaration of war on stablecoins by regional banks. The framing is accurate: Cari gives regulated deposit institutions a native on-chain form factor that competes with Tether and Circle without requiring any stablecoin-specific legislation to apply to them.

Timeline: Q3 2026 pilot with design partner banks validating the core token lifecycle (issuance, transfer, redemption). Customer-facing availability targeted for Q4 2026, initially for inter-bank money movement. The American Bankers Association has backed the testing process.

ZK Stack: The White-Label Rollup Framework​

Alongside Prividium, ZKsync is evolving its ZK Stack from a framework for deploying individual chains into an orchestrated system of public and private networks with native cross-chain connectivity. The pitch to institutions is white-label ZK rollup deployment — any bank or market infrastructure provider that wants its own blockchain (rather than shared tenancy on a public chain) can deploy a ZK Stack chain under its own brand and compliance stack, without building the zero-knowledge proof infrastructure from scratch.

Cross-chain connectivity in the ZK Stack model does not require external bridges. Applications deployed on ZK Stack chains can access liquidity and shared services across ZK chains and Ethereum natively, which eliminates one of the key systemic risks (bridge exploits) that has historically made enterprise blockchain architects nervous.

The evolution mirrors what happened to cloud infrastructure between 2010 and 2020: the movement from single-tenant hosted servers to shared public cloud to private cloud to hybrid architectures. ZK Stack is ZKsync's private cloud offering for the blockchain layer.

Airbender: The Proving Engine Underneath​

Behind both Prividium and ZK Stack sits Airbender, ZKsync's next-generation zero-knowledge proof system. ZKsync describes Airbender as "the fastest RISC-V zkVM in the world" and positions it as the universal proving standard for ZKsync chains.

The performance numbers matter for institutional viability. Airbender targets proof generation costs around $0.0001 per transfer, with proof speeds fast enough to support real-time transaction confirmation UX. At institutional transaction volumes — ZKsync's 2026 roadmap explicitly references "trillions of dollars in annual transaction volume" as the design target — proof generation throughput is not an abstract benchmark but an operational cost that shows up in settlement economics.

The RISC-V architecture choice is significant. It means Airbender can prove execution of programs written in any language that compiles to RISC-V — which includes most systems programming languages — rather than requiring specialized ZK-circuit implementations of each operation. For financial institutions with existing software assets in Java, C++, and Rust, that matters.

The Competitive Landscape: Where ZKsync's Bet Fits​

ZKsync's institutional pivot did not emerge in a vacuum. Multiple L2 ecosystems are making explicit identity bets in 2026.

Arbitrum has doubled down on DeFi TVL dominance and Stylus VM multi-language smart contracts, targeting the large developer base that wants Rust and C++ on Ethereum. Base (Coinbase) is optimizing for consumer onboarding — the simplest possible path from fiat to on-chain for retail users. Optimism's Superchain is focused on horizontal ecosystem growth through shared sequencing.

Polygon occupies the most direct competitive position relative to ZKsync's institutional bet. Polygon's "Open Money Stack" thesis — built around stablecoin settlement with Visa, Modern Treasury, and enterprise fintechs — targets the same regulated market. Visa chose Polygon as one of its five expanded stablecoin settlement networks in May 2026, a direct competitive overlap.

The key differentiator ZKsync presses is Prividium's default confidentiality. Polygon zkEVM uses ZK proofs for execution validity, but publishes transaction data publicly. For banks, that distinction is critical: a ZK proof that "this computation was correct" is not the same as a ZK proof that "this computation was correct, and no third party learned the participants or amounts." Prividium delivers both properties. Polygon delivers only the former.

Starknet offers a competing ZK-based architecture with a Stage 1 decentralization rating (stronger than ZKsync's current Stage 0). The decentralization gap is a genuine risk: ZKsync's Stage 0 status means users cannot permissionlessly exit funds without Matter Labs' sequencer operating correctly, which is a concern institutional risk teams will surface in evaluations. ZKsync's roadmap addresses sequencer decentralization but the work is ongoing.

The prior institutional blockchain attempts — Canton Network (Goldman-backed permissioned DeFi), Fabric (Hyperledger enterprise), Corda (R3 bank consortium) — collectively failed to reach production scale beyond pilot deployments. ZKsync's architecture makes a different bet: rather than a permissioned chain disconnected from public liquidity, Prividium connects institutional privacy with Ethereum's settlement layer and global liquidity. That interoperability argument is the core reason the Cari Network chose ZKsync over a purely permissioned alternative.

What This Means for Infrastructure Builders​

ZKsync's 2026 roadmap represents a concrete data point in the broader institutional L2 market: regulated financial institutions are moving beyond pilots to production timelines, and the winning infrastructure architecture appears to combine ZK-proof-backed verification with selective privacy and Ethereum settlement connectivity.

For developers and infrastructure teams building on or around these chains, the architectural shift creates new indexing and data requirements. Prividium deployments produce different event shapes than public DeFi — attestation lookups, custody-chain verifications, and settlement-finality polls dominate the traffic pattern rather than the swap-heavy profile of consumer DeFi chains. Institutional-grade RPC providers must align SLAs with settlement clearing windows measured in seconds, not the permissive degradation tolerances that consumer apps typically accept.

The Cari Network's Q3 2026 pilot will be an important stress test. If five regional banks with $600 billion in deposits successfully run tokenized deposit transactions through a ZK rollup and reach customer-facing launch in Q4, ZKsync's institutional pivot will have its first durable proof-of-concept — one that prior enterprise blockchain efforts never achieved.

The L2 market in 2026 is no longer a race to the same finish line. ZKsync has explicitly chosen a different race, and the starting gun has already fired.

BlockEden.xyz provides enterprise-grade RPC and indexing infrastructure across 27+ blockchain networks including Ethereum and its L2 ecosystem. As institutional adoption of on-chain infrastructure accelerates, explore our API marketplace for production-ready access to the chains that matter.

Sources:

Share on Twitter