a16z: The three major trends in artificial intelligence for 2026

Author: a16z crypto

Translation: Deep潮 TechFlow

This year, AI will undertake more substantive research tasks

As a mathematical economist, as early as January 2025, I found it difficult to get consumer-grade AI models to understand my workflow; however, by November 2025, I was able to give abstract instructions to AI models just like giving directives to PhD students… and sometimes they even returned some novel and correct answers. Beyond my personal experience, AI is being more broadly applied in research, especially in reasoning. These models not only directly assist in discovery but can also autonomously solve difficult problems such as the Putnam problem (perhaps the hardest university math exam in the world).

It remains uncertain which fields will benefit most from this research assistance approach and how exactly it will be implemented. But I expect that this year, AI research will promote and reward a new style of “jack-of-all-trades” research: one that emphasizes exploring the relationships between various ideas and can quickly infer from more hypothetical answers.

These answers may not be entirely accurate, but they can still guide research in the right direction (at least within certain topological structures). Ironically, this is somewhat like leveraging the power of models’ “hallucinations”: when models are “smart enough,” giving them an abstract space to stir their thoughts may still produce some meaningless results—but sometimes it can lead to breakthrough discoveries, much like humans working outside linear thinking or explicit directions, which can be the most creative.

Reasoning in this way requires a new AI workflow style—not just simple “agent-to-agent” interactions, but a complex collaboration pattern of “agent nesting agents.” In this mode, different levels of models assist researchers in evaluating early-stage proposals and gradually distill the essence. I have already been using this method to write papers, while others are conducting patent searches, inventing new forms of art, and (regrettably) discovering new attack methods for smart contracts.

However, to operate these nested reasoning agent combinations for research, better interoperability between models and a method to identify and appropriately compensate each model’s contribution are needed—and these issues might be addressed by blockchain technology.

—Scott Kominers (@skominers), a16z Crypto research team member, Harvard Business School professor

From “Know Your Customer” (KYC) to “Know Your Agent” (KYA): the shift in identity verification

The bottleneck of the agent economy is shifting from intelligence to identity verification. In financial services, the number of “non-human identities” now exceeds 96 times that of human employees—yet these “identities” remain “ghosts” that cannot access banking services.

The missing key infrastructure here is “Know Your Agent” (KYA). Just as humans need credit scores to obtain loans, agents also need cryptographic signatures as credentials to conduct transactions—these credentials link the agent to its entity, constraints, and responsibilities. Until this infrastructure is established, merchants will continue to block these agents at firewalls.

The industry that has built KYC infrastructure over the past decades now has only a few months to explore how to implement KYA.

—Sean Neville (@psneville), co-founder of Circle, architect of USDC; CEO of Catena Labs

Solving the “Invisible Tax” problem of open networks: economic challenges in the AI era

The rise of AI agents is imposing an “invisible tax” on open networks, fundamentally disrupting their economic foundation. This disruption stems from the increasing mismatch between the “context layer” and the “execution layer” of the internet: currently, AI agents extract data from ad-supported websites (the context layer), providing convenience to users but systematically bypassing the revenue sources supporting content (such as ads and subscriptions).

To prevent the gradual decline of open networks (and protect the diverse content fueling AI), we need large-scale deployment of technological and economic solutions. These might include next-generation sponsorship content models, micro-attribution systems, or other new funding mechanisms. However, existing AI licensing agreements have proven to be financially unsustainable, often only compensating a small portion of content providers’ lost revenue due to AI traffic diversion.

The network urgently needs a new technological-economic model that allows value to flow automatically. The key transition in the coming year will be from static licensing models to real-time usage-based compensation mechanisms. This involves testing and scaling relevant systems—possibly leveraging blockchain-supported nanopayments and complex attribution standards—to automatically reward entities that contribute information successfully completing tasks for AI agents.

—Liz Harkavy (@liz_harkavy), a16z Crypto investment team