Based on a16z's Annual Predictions, Which Crypto Projects Are Worth Watching?

Original Source: Stacy Muur

Original Compilation: Ken, ChainCatcher

In mid-December, a16z released its annual "Big Ideas 2026" report. The following is a watchlist compiled based on the key theses from their vision.

1. The Reshaping of Payments, Stablecoins, and Financial Rails

Thesis: Stablecoins will become the settlement layer for the internet, not just for cryptocurrency.

The stablecoin narrative has moved past the proof-of-concept stage. Today, their annual transaction volume has reached trillions of dollars. The question is no longer whether stablecoins work, but whether they can integrate seamlessly with the real-world financial system.

a16z's argument here is crucial: stablecoins are not just currency; they are an upgrade to the ledger. If stablecoins can coexist with existing systems, offering real-time settlement, programmability, and global reach, banks and fintech companies won't need to rewrite decades-old legacy software. This shifts the focus of value capture from issuing stablecoins to distribution, compliance, and integration.

Who Wins in 2026?

• Embedded payments.
• Card issuance (crypto neo-banks) and wallets.
• Bank-grade compliance + APIs.
• Global, programmable settlement.

Watchlist:

• Circle (USDC): Regulatory legitimacy + distribution + liquidity advantage. Watch: Payment APIs, bank integrations, inter-agent usage.
• m0: Zero-fee USDC issuance based on modular infrastructure. Worth watching: L2-native minting, stateless agents, browser-level stablecoin UX.
• Ether_fi: The hottest new banking contender for 2026; offers yield via tokenized treasuries, cashback, and seamless payments; bridges crypto and fiat growth.
• Plasma: Billed as the first Neobank built entirely around stablecoins, targeting global users (especially emerging markets) whose daily finances rely on USD-backed stablecoins.
• Stablecoin (Bridge) | Stripe Integration: On/off-ramp infrastructure that abstracts the crypto layer. Watch: Integration with regional payment rails.
• x402 Ecosystem: Watch: Progress in programmable payments "for Agents" and "by Agents" via native HTTP-level settlement.

2. RWA: Native Issuance, Not Just Tokenization

Thesis: Tokenization alone is not enough. To truly improve efficiency, the source of funding must move on-chain.

The first wave of real-world asset (RWA) applications primarily focused on tokenizing existing off-chain instruments (e.g., loans, treasuries, credit products) and distributing them to crypto users. While this improved accessibility, it largely preserved the old inefficiencies: opaque underwriting processes, high servicing costs, slow settlement, and fragmented liquidity. In many cases, tokenization was just a new wrapper for old processes.

a16z's core insight is that crypto's true advantage lies not in replicating traditional financial structures, but in reshaping the credit system from the source. When loans are originated on-chain, underwriting logic becomes programmable, servicing costs plummet, and risk can be priced and monitored in real-time. This is where crypto stops being just a distribution channel and starts becoming financial infrastructure.

Who Wins in 2026?

• On-chain underwriting
• Transparent risk pricing
• Compliant credit mechanisms
• Deep liquidity (often via Perps)

Watchlist:

• Centrifuge: On-chain credit rails for RWAs; watch: Institutional deal flow and default performance.
• Blackrock (BUIDL): Tokenized money market fund on Ethereum; watch: Inflows into tokenized treasuries and TradFi adoption pace.
• Maple: Institutional lending with underwriter governance; watch: Credit line expansion and net yield vs. default risk.
• Plume: Composable credit infrastructure for programmable debt; watch: Custom underwriting logic and DAO use cases.
• Pendle: Yield stripping for tokenized yields; watch: PT/YT adoption for RWAs in treasuries and private credit.
• Ondo: Tokenized treasury and credit funds; watch: Compliant USDC-to-RWA pipeline and L2 expansion.
• Backed: Regulated ETF and bond token wrappers; watch: DeFi-native compliant rails.

3. The Internet Becomes a Bank (Smart Agents & Payments)

Thesis: As AI agents begin to transact autonomously, payments can no longer be an external system bolted onto applications. They must be internet-native: instant, programmable, and fully automated.

The key shift is from user-driven execution to intent-driven execution. Agents no longer click buttons or approve invoices; they identify conditions, fulfill obligations, and autonomously trigger actions. In this model, traditional payment processes (invoicing, batch processing, reconciliation, settlement windows) are no longer operational details but structural bottlenecks.

Blockchains introduce a different model. Smart contracts already enable global final settlement in seconds. Emerging underlying technologies push this further, making value transfer more responsive and composable: an agent can pay another agent for data, computation, or service immediately upon task completion, with rules embedded directly in code, enforced without intermediaries. Money ceases to be a separate operational layer and starts to behave like network traffic, which the internet can handle natively.

Who Wins in 2026?

• Agent-native identity.
• Programmatic payment rails.
• User experience without "human" intervention.

Watchlist:

• Catena: Identity and compliance infrastructure for AI Agents; establishing "Know Your Agent" (KYA) standards. Watch: Agent onboarding, enterprise-grade integration.
• Nevermined: Data marketplace infrastructure for autonomous agents; enables permissioned agent-to-agent payments for access, computation, and services. Watch: Executable data licenses, agent monetization flows.
• KiteAI: AI-native agents with embedded payments and real-world task completion capabilities. Watch: Intent-based automation, real economic throughput.
• ASI: Open ASI + blockchain fusion layer; cross-agent collaboration and compute monetization. Watch: Agent settlement standards, autonomous service economy.
• EigenCloud (via EigenAI): Deterministic AI agents for decentralized policy execution and autonomous protocol operations. Watch: Intent-driven finance, composable economic agency.
• Fetch: Multi-agent collaboration protocol enabling decentralized compute + services. Watch: On-chain task execution, measurable Agent GDP.
• x402 Implementation: Agent settlement, protocol-level payments, inter-agent transactions.

4. Privacy as the Primary Moat

Thesis: Privacy leads to user lock-in. Public blockchains commoditize users.

The core insight of a16z's privacy theory is simple: block space has become interchangeable, but confidentiality has not. Performance, fees, and throughput are no longer durable differentiators. If everything is public, users can migrate freely, liquidity can be bridged instantly, and applications can compete in a zero-profit environment. Privacy breaks this symmetry.

Once users, institutions, or applications commit sensitive state—balances, strategies, counterparties, identities, metadata—into a privacy-preserving environment, switching costs emerge naturally. This creates a privacy network effect: the more activity within a private domain, the more value stays there, and the greater the risk of leaving due to boundary information leakage.

Who Wins in 2026?

• Private execution environments
• Zero-knowledge-based data access control
• Privacy-by-default, not as an add-on

Watchlist:

• Aztec: Private smart contracts + ZK-native Rollup. Watch: Developer traction, private DeFi primitives.
• Nillion: Decentralized MPC for private computation; institutional data custody use cases are the breakout signal.
• Arcium: Confidential compute layer in the Solana stack; watch: Performance scaling and Solana-native integration.
• Aleo: ZK cloud compute platform with native credits; enterprise zkCloud usage is the north star metric.
• Walrus and Seal: Key components of the Sui stack, enabling fully on-chain data and privacy.
• Payy_link: Privacy-preserving stablecoin wallet, combining cryptographic privacy with utility, enabling gasless, fully private, and optionally compliant sending/receiving of stablecoins like USDC.
• Zcash: Shielded transfers via ZK-SNARKs; the launch of Halo 2 and expansion of programmable privacy are key.
• Monero: Default privacy L1 with ring signatures; robustness under surveillance pressure remains its core moat.

5. Security: From "Code is Law" to "Spec is Law"

Thesis: Audits are not enough. Runtime enforcement will become standard practice.

The past two years have made one thing clear: audit failures are not due to auditor incompetence, but because audits are static, local, and fundamentally incapable of capturing dynamic systems. Today, protocols operate in adversarial environments shaped by Maximal Extractable Value (MEV), composability, oracle latency, and incentive-driven edge cases that often only emerge post-deployment, frequently under extreme market conditions.

a16z's conclusion (and one now widely accepted in the industry) is that security must move up a level: from checking if the code is correct, to enforcing that system-level invariants are absolutely unbreakable. This marks a shift in security strategy from "code is law" to "spec is law," where protocols formally define which rules must always hold (e.g., collateral limits, value conservation, solvency conditions, ordering constraints) and continuously enforce these properties, not just at deployment.

Who Wins in 2026?

• Runtime guardrails
• Formal specifications
• AI-assisted verification
• Continuous monitoring

Watchlist:

• OpenZeppelin: Runtime rule enforcement and upgrade security platform with deep protocol integration.
• Trailofbits: Advanced security research firm pivoting to real-time invariant enforcement via tooling.
• SpearbitDAO: Research-driven audit collective launching continuous audit pipelines (vs. static snapshots).
• Cyfrin: Smart contract security firm building educational pipelines and formal-verification-first tooling; growing influence in modular L2 ecosystems.
• Immunefi: Unified on-chain platform for bug bounties, audits, and AI-driven threat detection to protect crypto projects from exploits.
• Runtime monitoring startups.

6. Prediction Markets at Scale

Thesis: Prediction markets evolve from niche betting venues to the internet's real-time information infrastructure.

The core shift a16z points to is not merely "more markets" or higher volume, but a reshaping of prediction market structure. As block space costs fall, oracle performance improves, and UX friction decreases, markets cease to be occasional events tied to elections or sports and begin to function as a continuous signal extraction layer. Everything becomes tradable: macro data releases, protocol upgrades, regulatory votes, corporate actions, even the probabilistic outcomes of long-tail events.

Liquidity fragments across thousands of micro-markets, but price discovery improves because information is no longer constrained by polls, surveys, or centralized analysts.

Who Wins in 2026?

• Prediction markets that can list everything.
• AI agents trading continuously.
• Decentralized and probabilistic resolution.

Watchlist:

• Polymarket: Leading decentralized prediction platform on Polygon; over $1B monthly volume, dominant in crypto.
• Kalshi: CFTC-regulated US exchange; over $1.3B monthly volume and top app downloads in legal markets.
• FractionAI: Hailed as the first AI agent prediction market, where agents compete with live portfolios in a transparent on-chain environment.
• Opinion: High-growth event exchange with over $700M monthly notional volume, breaking through in crypto-native wagering.
• Myriad Markets: Decentralized event market infrastructure; over $10M volume, active on-chain betting across niches.

7. zkVMs and Verifiable Computation

Thesis: Proofs leave the blockchain and enter the cloud.

The key inflection point a16z identifies is not a slight speed-up in zero-knowledge proofs, but that zkVMs are crossing a threshold where proving general computation is no longer a blockchain-specific luxury but a viable system primitive, thanks to dramatic reductions in proving overhead (from ~1,000,000x to ~10,000x), GPU-native provers, and memory footprints suitable for real production.

Together, they unlock a regime where ordinary CPU workloads (cloud jobs, backend services, financial models, ML inference, legacy enterprise code) can be executed once and verified anywhere, transforming trust assumptions about cloud providers, data pipelines, and off-chain execution into cryptographic guarantees rather than contractual promises, and creating a world where correctness, not reputation, is the default security model for distributed systems.

Watchlist:

• RiscZero: zkVM with general Rust compute capability; credible lead in cloud-compatible proving.
• Succinct: SP1 zkVM (RISC-V/LLVM based) for arbitrary Rust/LLVM code; focuses on fast verification, light clients, and off-chain data feeds, with GPU-native proving (e.g., real-time proving on 16 GPUs), precompiles for efficiency, and recursion for on/off-chain use.
• Brevis_zk: zk coprocessor for querying on/off-chain data; designed for generating modular proofs inside applications.
• Axiom_xyz: ZK coprocessor for verifiable computation over on/off-chain data; supports arbitrary expressive computation off-chain (e.g., historical queries) and verifies ZK proofs on-chain.
• ZKML Stack Implementations: Frameworks for verifiable ML inference via ZK proofs/SNARKs (e.g., ZKML, DSperse, JSTprove); optimize circuits for production ML workloads (e.g., GPT-2, neural nets), reducing overhead (5-22x faster proving/verification).

8. Wealth Management Moves On-Chain

Thesis: Active, personalized wealth management becomes ubiquitous.

As assets—including yield-bearing cash, public equities, private credit, and illiquid alternatives—tokenize natively, rebalancing ceases to be an event and becomes a continuous process. This process is executed by smart contracts responding to real-time changes in interest rates, volatility, and risk premiums, not quarterly meetings or advisor workflows.

AI-assisted asset allocation engines increasingly play a co-pilot role, translating user constraints into executable strategies; while DeFi primitives like automated vaults, layered yield products, and permissioned RWA pools provide the mechanism depth to implement these strategies without sacrificing compliance or capital efficiency. The result is not a "robo-advisor" but programmable wealth, where portfolios self-adjust along risk curves, rotating from tokenized treasuries to credit, from Beta to Carry, from liquid to illiquid, with settlement, custody, and reporting abstracted away.

What Wins?