A16z: AI agents and on-chain finance are about to reshape everything

A16z has predicted a big pivot in 2026, as stablecoins surge, AI agents proliferate, and on-chain financing challenges outdated banking infrastructure.

A16z crypto, a venture capital fund of Andreessen Horowitz (a16z), said it expects several crypto-related themes to gain importance in 2026, citing developments in stablecoins, real-world asset tokenization, banking infrastructure, AI agent authentication, privacy technologies and prediction markets.

In its latest post, the company said stablecoins processed an estimated $46 trillion in transaction volume last year, even surpassing major payment networks. But connecting digital dollars to existing financial rails remains an unsolved problem.

Major crypto themes for 2026

According to the report, new startups are building on-ramps that link stablecoins to local payment systems, QR-based regional networks, real-time settlement rails and merchant tools, making it easier for digital dollars to circulate in everyday payments. A16z said these integrations could support new use cases such as instant cross-border payroll and direct adoption by merchants without bank accounts.

The report also noted growing interest among banks, fintech companies and asset managers in bringing traditional assets on-chain, but said many tokenization efforts still replicate existing financial structures. It spoke of synthetic instruments such as perpetual futures as easier to implement and capable of deeper liquidity, adding that “perpification versus tokenization” will remain a key question as more assets come on-chain.

Stablecoin issuance also grew in 2025, and A16z predicted that 2026 would see more on-chain issuance of credit products instead of off-chain issuance that would later be converted into tokens. The company said banks continue to rely on old core systems built decades ago and argued that stablecoins, tokenized deposits, tokenized treasuries and on-chain bonds allow institutions to launch new products without replacing aging infrastructure.

In its vision for automation, the report states that a shift to intent-based systems and AI agents will require payments that run at internet speeds, supported by programmable settlement tools such as x402. This could allow value transfer to become a native network function rather than a separate operational layer.

Prediction Markets, DeFi and LLM Oracles

The report also predicted broader access to asset management tools as tokenized assets, stablecoins and DeFi allocation mechanisms enable faster portfolio adjustments, including access to private market assets that have traditionally been limited to institutional investors.

You might also like:

In terms of identity, A16z said that there are now more non-human agents than human employees in the financial services industry, and argued that cryptographically signed credentials, described as ‘Know Your Agent’, will be required for agents to conduct transactions. The report states that AI systems are increasingly being used for research tasks and predicted that multi-agent workflows will require attribution and compensation mechanisms that blockchain systems could support.

A16z believes that the rise of AI agents has disrupted the economic model of the open web by extracting information without supporting ad-based revenue, adding that real-time usage-based compensation systems, possibly using crypto micropayments, may be necessary.

On the privacy front, the venture fund said chains with built-in confidentiality could experience stronger network effects, as moving between private and public environments compromises exposure to metadata. The report also predicted an increased focus on decentralized, quantum-resistant messaging networks and described “secrets-as-a-service” as a necessary framework for enforcing data access rules across the chain.

According to the report, recent DeFi exploits show that current security practices remain largely on a case-by-case basis, and that the industry will need to shift to proving system-wide properties and enforcing them at runtime. It also noted that the prediction markets are expected to list many more contracts, which could raise questions about how outcomes can be determined, and that new decentralized governance mechanisms and LLM-based oracles will be proposed to help resolve disputed events.