a16z's latest research: Three core trends of AI + Crypto in 2026

Original Title: AI in 2026: 3 trends

Original Author: a16z crypto

Original Compiler: Ken, Chaincatcher

1. This year, AI will take on more substantial research work

As a mathematical economist, back in January 2025, I found it difficult to get consumer-grade AI models to understand my workflow; but by November, I was able to give abstract instructions to the models as if I were guiding a PhD student… and sometimes they provided novel and correct answers. Beyond my personal experience, AI is being more widely applied in research fields—especially at the reasoning level, where models are directly assisting in discovery (innovation points) and even autonomously solving difficult problems from the Putnam Mathematical Competition (one of the hardest university math exams in the world).

It is still unclear which fields this type of research assistance will be most effective in and how exactly it will work. However, I anticipate that AI research this year will give rise to and reward a new type of "polymathic" research style: one that values the ability to speculate on the connections between concepts and the ability to quickly infer conclusions from uncertain answers.

These answers may not be entirely accurate, but under certain topological structures, they can point in the right direction. Ironically, this is somewhat akin to harnessing the power of "model hallucination": when models are smart enough, giving them space for abstract exploration may produce nonsense, but sometimes it can also open a door to discovery—just as humans tend to be most creative in nonlinear, ambiguous instruction states.

This reasoning approach requires a brand new AI workflow—not just interactions between agents, but more of a "agent encapsulating agent" model—where multi-layered models can help researchers evaluate the outputs of early models, filtering through layers to discern truth from falsehood. I have been using this method to write papers, while others use it for patent searches, inventing new art forms, or (unfortunately) uncovering vulnerabilities in new smart contracts.

However, running this "encapsulated reasoning agent cluster" for research requires better interoperability between models, as well as a mechanism to identify and fairly compensate each model's contributions—and cryptocurrency can help solve both of these issues.

2. Moving from "Know Your Customer" to "Know Your Agent"

The bottleneck in the agent economy is shifting from "intelligence" to "identity." In the financial services sector, the number of "non-human identities" is already 96 times that of human employees—yet these identities are ghostly figures not covered by banks.

The missing key fundamental element here is KYA: Know Your Agent.

Just as humans need credit scores to obtain loans, agents also need cryptographically signed credentials to transact—binding agents to their principals, constraints, and liability. Until this credentialing system is established, merchants will continue to keep agents behind firewalls.

The industry that took decades to build KYC infrastructure now has only a few months to get KYA right.

3. We must address the "invisible tax" on open networks

The rise of AI agents is imposing an invisible tax on open networks, fundamentally undermining their economic foundations. This undermining stems from the increasing misalignment between the internet's "content layer" and "execution layer": current AI agents extract data from websites that rely on advertising (content layer) to provide convenience to users, while systematically bypassing the revenue sources that support content creation (such as advertising and subscriptions).

To prevent the erosion of open networks (and to protect the diverse content that feeds back into AI development), we need to deploy large-scale technological and economic solutions. This may include next-generation sponsored content, micro-attribution systems, or other new financing models. Existing AI licensing agreements have proven to be merely unsustainable stopgap measures, often compensating content providers for only a small fraction of the revenue lost due to AI siphoning traffic.

The network needs a brand new "tech-economy" model that allows value to flow automatically. The key transformation in the coming year lies in shifting from static licensing to real-time, usage-based compensation mechanisms. This means we need to test and scale new systems—potentially leveraging blockchain-enabled micropayments and complex attribution standards—to automatically reward every entity that contributes information for the successful execution of tasks by agents.