On-Chain Privacy: From "Optional" to "Indispensable"

Author: ChainUp Investment

1. Introduction

By 2025, on-chain privacy experienced a major re-pricing event. Notably, driven by a resurgence of privacy awareness within the industry and significant advances in cryptographic technology, Zcash achieved substantial price discovery. These technological advancements include zero-knowledge proofs (ZKPs), multi-party computation (MPC), trusted execution environments (TEE), and fully homomorphic encryption (FHE).

ZKPs: A method of proving the validity of a statement without revealing any information beyond its validity, enabling users to publicly share proofs of knowledge or ownership without disclosing details.

MPC: A cryptographic protocol involving multiple parties jointly computing data by splitting it into “secret shares.” No single party can see the complete data.

TEE: A hardware-based solution. It is a secure “black box” inside the processor used to isolate data during processing.

FHE: An encryption scheme that allows direct computation on encrypted data without decryption.

The market has shifted from “anonymity” to “confidentiality,” which is a functional necessity in transparent blockchains.

In Q4 2025, interest in on-chain privacy surged, according to Dexu.

1.1. Privacy Paradox

The history of privacy-focused cryptocurrencies dates back to 2012 when Bytecoin launched CryptoNote, which introduced ring signatures later adopted by Monero in 2014. In short, privacy in cryptocurrencies is not a new concept, but in early cycles, privacy coins were largely driven by ideological pursuits or means of evasion, serving as channels for bad actors to escape surveillance. The on-chain privacy dilemma in previous years can be attributed to three main factors: immature technology, fragmented liquidity, and regulatory hostility.

Historically, cryptographic techniques faced high latency and cost inefficiencies. Today, developer tools like Cairo (zkDSLs) and back-end “booms” like Halo2 enable mainstream developers to utilize ZKPs. The trend of building zkVMs (zero-knowledge virtual machines) on standard instruction sets like RISC-V makes the technology scalable and composable across various applications. MPC is no longer just for splitting private keys; with MP-SPDZ, it supports arithmetic circuits (addition/multiplication) and Boolean circuits (XOR/AND), enabling general-purpose computation. Advances in GPUs, such as H100 and Blackwell B200, now support confidential computing, allowing AI models to run within TEE. The biggest bottleneck of FHE—bootstrapping delay (the time to “refresh” noise in encrypted computations)—has decreased from about 50 milliseconds in 2021 to under 1 millisecond in 2025, enabling real-time deployment of FHE-based smart contracts.

Iterative improvements in zkVMs and performance are evident, as shown by Succinct and Brevis.

Moreover, privacy is often isolated within specific blockchains, forcing users to leave active ecosystems across chains to achieve anonymity, which incurs high transaction fees and opportunity costs. Today, protocols like Railgun can directly integrate with DeFi applications, providing privacy as a shield against front-running and MEV extraction. Protocols such as Boundless, Succinct, Brevis, and similar offer ZKP-as-a-Service, while Arcium and Nillion help build privacy-preserving applications using MPC, and Phala and iExec perform confidential computations within TEE without leaving their blockchains. Finally, Zama and Octra enable native on-application FHE processing.

Railgun TVL, source: DefiLlama

Initially, blockchains required transparency to gain legitimacy. True builders had to distance themselves from hackers, money launderers, and other bad actors. In this environment, privacy features were quickly viewed as tools for dishonest participants. Projects like Tornado Cash, while favored by privacy-conscious users, placed those users at risk of funds being mixed with illicit activities, making it impossible to prove innocence. This led to regulatory crackdowns. Exchanges froze funds from mixers and delisted suspicious privacy tokens to pursue operational licenses. Venture capital and institutional funds avoided holding such assets due to compliance concerns. On-chain privacy became a “criminal” feature in the industry. Now, sanctions on Tornado Cash have been lifted. The industry has coalesced around the concept of “compliant privacy,” designing “viewable data” that allows users to decrypt their own transaction sources by providing “view keys” to auditors or regulators, as seen in both Tornado Cash and Zcash.

Impact of sanctions on Tornado Cash fund flows, source: Dune

2. Current Privacy Technology Use Cases

Early setbacks do not mean privacy is unimportant. Ask yourself: “Do you want your actions of buying coffee today to expose your entire 10-year investment history?” Most would say no, but that is exactly what current blockchain setups do. As crypto legislation advances and more institutions participate, these new players are re-evaluating this issue. Fortunately, by 2025, the adoption of privacy tech is more driven by functional utility than ideology.

2.1. Shielded Transactions

Using “viewable data” design, Zcash’s shielded supply increased from 12% at the start of 2025 to approximately 29% now. The demand stems from multiple factors, such as increased speculative interest in ZEC tokens and a natural desire to shield transactions from the public. The mechanism for shielded transactions is called the Commitment-Nullifier scheme, where senders can deposit shielded funds into a pool, and the network verifies the deposits using ZKPs to prevent double-spending, creating a new shielded wallet for the recipient.

Zcash shielded ZEC supply, source: ZecHub

One of the fastest-growing areas is crypto-neobanks, actively exploring privacy features for their users, such as Fuse, Avici, and Privily. Although these protocols use different methods to obscure on-chain transactions.

2.2. High-Performance Execution Environments

Based on total value locked (TVL), ZK-2 layer networks grew by 20% in 2025, offering significantly cheaper execution environments compared to Ethereum Layer 1. These Layer 2s bundle all transactions into a small data blob, send it to a sequencer to generate proofs, and submit them to the base layer for verification.

Application trends in major ZK-2 layer networks: TVL changes, source: DefiLlama

Today’s ZK solutions provide comprehensive built-in privacy features, such as privacy smart contracts on Aztec and ZKsync Interop, which unifies liquidity between ZK chains and Ethereum.

2.3. MEV Protection

One of the most common “hidden” use cases for privacy is to prevent MEV (Maximal Extractable Value). The transparent nature of blockchains allows frontrunning bots to see transactions in the mempool before confirmation and perform sandwich or arbitrage trades for profit. Flashbots SUAVE decentralizes block construction by encrypting the mempool, keeping transactions confidential until the block builder commits to including them. Unichain also introduces TEE-based block construction to ensure that transactions on Layer 2 cannot be frontrun.

Percentage of transactions protected by Flashbot Protect, source: Dune

2.4. Other Use Cases

Beyond the main use cases, developers are actively exploring on-chain privacy implementations to optimize applications and improve user experience.

Order books: Whales like James Wynn and Machi Big Brother on Hyperliquid often face liquidation hunting. While Hyperliquid’s founders believe transparency provides fair competition for market makers and tighter spreads, large traders see the risk of frontrunning or adverse trades as a significant negative. This creates opportunities for Aster, which offers hidden orders and plans to launch new shielded modes (Shield Mode) by 2026.

Identity: Activities such as new bank account applications and ICOs require identity verification. Protocols like idOS allow users to upload KYC once and seamlessly reuse it across compliant protocols. zkPass helps provide Web2 credentials without revealing details, while World ID uses iris hash proofs to verify identity, and ZKPassport confirms user identity without leaving data on the device.

SEC Chair Paul Atkins stated that many ICO types should not be considered securities and thus fall outside SEC jurisdiction. His stance may soon lead to increased ICO fundraising, boosting demand for crypto KYC solutions.

Cross-chain bridges: Historically, cross-chain bridges have been vulnerable to exploits, such as Ronin Bridge and Multichain, which were hacked for $624 million and $126 million respectively due to private key leaks. ZK-based cross-chain bridges minimize trust assumptions; once proofs are generated and verified, they provide immediate certainty and are scalable with transaction volume. Polyhedra Network connects over 30 chains via zkBridge and can serve as an “DVN” within LayerZero V2 stacks.

AI: ZK can help verify that outputs are generated based on expected inputs and processed by specific models. Giza enables trustless agents to execute complex DeFi strategies based on verified outputs. Phala uses Intel SGX enclaves to securely store private keys and sensitive data within AI agents.

3. Core DeCC Ecosystem Classification

On-chain privacy generally refers to decentralized confidential computing networks (DeCC). Although the market tends to classify protocols based on underlying privacy tech, each privacy stack involves trade-offs, and increasingly, protocols adopt hybrid approaches. Therefore, it’s best to categorize them as privacy blockchains, privacy middleware, and privacy applications.

Core DeCC Ecosystem Classification

3.1. Privacy Blockchains

The “privacy blockchain” category includes layer-1 and layer-2 networks where privacy mechanisms are embedded into consensus or execution environments. The main challenge for these networks is “cross-chain barriers.” They must attract users and liquidity from established blockchains; without killer apps making migration economically viable, this is extremely difficult. Privacy layer tokens are often allocated a “layer-1 network premium” as they serve as security collateral and gas tokens.

3.1.1. Legacy and Evolution of Layer-1 Privacy

Zcash has long been positioned as a Bitcoin-like network with privacy features. It employs a dual-address system allowing users to switch between transparent and shielded transactions, with “view keys” to decrypt transaction details for compliance.

The protocol is transitioning from proof-of-work (PoW) consensus to a hybrid model called Crosslink, which will incorporate proof-of-stake (PoS) elements by 2026, providing faster certainty than Satoshi’s original probabilistic finality. After the halving in November 2024, the next halving is expected around November 2028.

Meanwhile, Monero maintains its default privacy approach, using ring signatures, stealth addresses, and ring confidential transactions (RingCT) to enforce privacy on every transaction. This design choice led most exchanges to delist XMR in 2024. Additionally, Monero experienced several Qubic hash rate attacks in 2025, causing up to 18 blocks of reorganizations and erasing approximately 118 confirmed transactions.

Secret Network is a TEE-based privacy layer built on Cosmos SDK since 2020, with view keys for access control. It positions itself both as an independent chain and as a provider of TEE-as-a-Service for EVM and IBC chains. The team also explores confidential AI computing and integrating threshold FHE into the network.

Canton Network, supported by Goldman Sachs, J.P. Morgan, Citi Ventures, Blackstone, BNY Mellon, Nasdaq, and S&P Global, is a layer-1 chain aiming to introduce trillions of dollars in RWA (real-world assets) via a unique privacy feature called Daml Ledger Model. Parties connected to its subnet can only see a subset of the ledger, enabling verification only among involved parties, with outsiders unaware of transaction existence.

Aleo is a ZK layer-1 network using a proprietary Rust-based language Leo, compiling code into ZK circuits. Users generate proofs off-chain (or pay miners to do so), then only submit encrypted proofs to the network.

Inco positions itself as an FHE layer-1 network, providing FHE-as-a-Service across chains via cross-chain bridges and messaging protocols. This enables deep liquidity without building DeFi from scratch.

Octra is a high-performance FHE layer-1 network, developing its own cryptography called Hypergraph FHE (HFHE), allowing parallel processing during computation, achieving a peak throughput of 17,000 TPS on its testnet.

Mind Network leverages EigenLayer and other restaking protocols to secure its FHE validator network. It aims to create an end-to-end encrypted internet (HTTPZ) and enable AI agents to handle encrypted data.

3.1.2. The ZK Layer-2 Networks

ZKsync has evolved from simple scaling to implementing comprehensive solutions like Prividium, ZKsync Interop, and Airbender. Prividium enables private transactions for companies while still using Ethereum for final settlement. Airbender is a high-performance RISC-V zkVM prover capable of generating ZK proofs in sub-second time. ZKsync Interop allows users to provide collateral on ZK chains and borrow assets on Ethereum.

Starknet uses STARKs (Scalable Transparent ARguments of Knowledge) for high-throughput scaling and features native account abstraction. Each Starknet account is a smart contract capable of executing private transactions. The team also proposes a ZK layer-2 on Zcash called Ztarknet, leveraging Zcash’s privacy.

Aztec operates as a native privacy layer-2 on Ethereum, using a UTXO-like note system for encrypted data and an account-based system for public data. It relies on Noir-based architecture with client-side proofs or PXE (privacy execution environments), where users generate ZK proofs locally and submit them to the network.

Midnight, a partner chain of Cardano, uses Cardano SPOs for security while running its own execution layer. It is a ZK layer-1 network built with TypeScript and selective disclosure features, staking with ADA, and using unshielded NIGHT tokens for governance and gas (DUST). Default-shielded DUST tokens serve as gas tokens.

Phala relies on Intel SGX enclaves to protect privacy. It has shifted toward AI co-processors, allowing AI agents to run within TEEs, managing private keys and sensitive data, collaborating with Succinct and Conduit to migrate from Polkadot parachains to Ethereum Layer 2.

Fhenix is Ethereum’s first FHE-based Layer-2, bringing encrypted computation into the Ethereum ecosystem. Transactions on this chain are protected against MEV because inputs are encrypted in the mempool.

3.2. Privacy “Middleware”

“Privacy middleware” protocols operate on a proof-as-a-service (PaaS) model, providing computational capabilities for proof generation, encryption, or verification. This space is highly competitive in latency, cost efficiency, and network support.

Boundless, incubated by RISC Zero, is a decentralized ZK proof marketplace, allowing any blockchain or application to outsource proof computation to Boundless.

Succinct Labs, a direct competitor to Boundless, positions itself as a high-performance proof network. It adds dedicated circuits for common tasks like hashing and signatures to its zkVM (SP1), making proof generation faster and cheaper.

Brevis acts as a ZK co-processor, enabling smart contracts to query historical blockchain data trustlessly. Now extended via Pico to a general zkVM, it can be pre-compiled for heavy workloads or integrated as dedicated circuits.

Arcium offers a configurable MPC solution serving any chain-based application, though it uses Solana for staking, slashing, and node coordination.

Nillion provides high-performance MPC services, with Nil Message Compute (NMC) and Nil Confidential Compute (nilCC) enabling shard data to be computed without message exchange, maintaining security within TEEs.

iExec RLC has been a long-standing DePIN protocol since 2017, providing cloud computing resources. It now emphasizes TEE-based confidential computing, enabling AI model training or inference without data leakage, supporting chains like Ethereum and Arbitrum.

Marlin has undergone a major transformation from a blockchain CDN to a confidential computing layer (Oyster) and built a ZKP marketplace (Kalypso) on top.

Zama is a leading FHE protocol building fhEVM, TFHE-rs, and Concrete, used by protocols like Fhenix and Inco. It also offers FHE-as-a-Service on existing public blockchains and plans to integrate FHE into zkVMs after acquiring Kakarot.

Cysic develops hardware (ASICs) to accelerate ZKP generation, reducing proof time from minutes to milliseconds. Users can request proof generation via ZK Air (consumer-grade) or ZK Pro (industrial-grade ASICs).

3.3. Privacy Applications

This is the largest category within privacy blockchains and middleware, with only a small subset listed here. These protocols leverage ZK, MPC, TEE, or MPC to enhance user experience. Successful applications abstract away privacy complexities and deliver truly market-fit solutions.

Tornado Cash was the pioneering decentralized, immutable mixer. It was sanctioned by the US Treasury in 2022 but had sanctions lifted in early 2025. Nonetheless, it remains a high-risk tool for compliant entities.

Railgun is well-regarded, endorsed by Vitalik Buterin. It integrates users’ “vaults” with DeFi protocols like Uniswap and Aave, offering voluntary shielded transactions beyond Tornado Cash. Its shielded assets are about 20% of Tornado’s, but it’s widely seen as a potential competitor.

World (formerly Worldcoin) uses iris scans to establish “personhood proofs,” with biometric data encrypted and only ZKPs sent to the network. World ID is an effective tool for distinguishing humans from AI.

zkPass employs third-party TLS handshakes to generate proofs of personal identity and media data, enabling access to gated apps without revealing private info.

Privy allows users to log into dApps seamlessly via email or Web2 accounts, creating MPC wallets with keys split between user devices and secure servers. This eliminates cumbersome mnemonic backups and greatly improves UX.

Aster collaborates with Brevis to build its Aster Chain, offering privacy transactions on top of existing hidden order protocols. Its roadmap indicates a launch in Q1 2026.

Malda is a unified liquidity lending protocol utilizing Boundless proofs to manage cross-chain lending positions.

Hibachi offers high-frequency decentralized perpetual trading, using Succinct proofs to verify off-chain CLOB orders on-chain.

Giza introduces machine learning into smart contracts, enabling them to execute complex DeFi strategies based on verified AI outputs. Giza allows trustless execution of AI-driven strategies on-chain.

Sentient is an AI-specific Layer-1 network (supported by Polygon CDK), aiming to create an open AGI platform with contributor rewards. AI owners upload models, which are encrypted and fingerprinted to verify outputs. It also develops Sentient Enclaves, leveraging AWS Nitro Enclaves for confidential AI model computation, shielding prompts and internal states from nodes.

4. Current Trends and Future Outlook

4.1.1. Rise of Privacy Middleware

We are witnessing a shift from monolithic privacy chains to modular privacy layers. Protocols no longer need to migrate to privacy blockchains but can deploy on any established chain (like Ethereum or Solana), accessing privacy services via smart contracts, thus lowering barriers. As demand for privacy features and industry privacy awareness grow, privacy middleware benefits most, as running heavy confidential compute frameworks locally is often economically unfeasible for many startups.

Requests and completions proofs on Succinct, source: Dune

4.1.2. Hybrid Solutions

Current privacy tech has limitations: ZKPs cannot compute on encrypted data; MPC may face latency issues with many participants; TEE can be compromised via fault injection or side-channel attacks if attackers gain physical hardware access; FHE computations can be slow, with increased noise risking data corruption. Therefore, protocols increasingly adopt hybrid approaches or design specialized hardware to optimize performance.

4.1.3. Confidential and Verifiable AI

Morgan Stanley estimates global AI-related capital expenditure at $3 trillion. As AI demand expands toward 2026, confidential and verifiable AI becomes a major trend in 2025, expected to scale further in 2026. Confidential model training on sensitive data like healthcare and financial records could mark another milestone in decentralized AI.

5. Summary

The era of privacy tokens without “view keys” may be ending. The industry bets that “selective disclosure” will be accepted as a reasonable compromise. If regulators later reject this approach, networks might be forced into “regulated permissioned chains” to maintain anonymity.

Maturity in privacy tech is key to unlocking “trillions of dollars” in traditional finance. Bonds, securities, and corporate payrolls cannot exist on transparent chains. As these protocols prove their robustness in 2025, the first major “privacy RWA” pilots are expected to launch on one of the aforementioned networks in 2026.

Google Trends for “blockchain privacy” over the past five years, source: Google

While interest in blockchain privacy may temporarily cool, demand at the application layer is expected to steadily grow, significantly improving user experience and attracting a broad non-crypto-native audience. This marks the moment when on-chain privacy shifts from “nice to have” to “indispensable.”