AI Agents Versus Human Traders: Nansen’s CEO Says the Contest Ends in Two Years
AI trading agents will overtake human traders in crypto markets within about two years. That’s the call Nansen CEO Alex Svanevik made publicly on July 28 — and his firm is now restructuring its entire product line around it.
Svanevik said Nansen is rebuilding its onchain analytics infrastructure specifically to serve autonomous agents rather than human analysts.
Also Read: Can €10 Billion Close Europe’s AI Compute Gap? Brussels Just Placed the Bet
It’s one of the clearest public commitments yet from a major crypto-data firm to a single idea: that AI agents, not people, will soon set prices and execute most digital-asset trades.
What AI Trading Agents Actually Are
Svanevik publicly stated that Nansen is rebuilding its onchain analytics infrastructure specifically to serve autonomous agents rather than human analysts. An AI trading agent is a software program that perceives market data, reasons about it using a language model or reinforcement-learning system, and executes trades without human approval on each transaction.
Unlike algorithmic trading bots that follow fixed coded rules, modern AI agents can adjust their strategies dynamically, interpret unstructured information such as governance proposals or protocol announcements, and chain together multiple actions across different protocols in a single session.
The distinction matters because fixed bots have existed in crypto since 2017. What Svanevik is predicting is qualitatively different: agents that can read an onchain governance forum, model the likely outcome of a vote, and position ahead of it.
That capability requires the kind of contextual reasoning that only large language models have demonstrated, and only at scale since 2024.
Nansen’s All-In Pivot To AI Trading Agents
Nansen built its reputation as a human-facing analytics platform. Its wallet-labeling system, which tags onchain addresses with identities like “smart money” or “exchange hot wallet,” became a standard reference for traders trying to follow informed capital flows.
Svanevik’s restructuring wager is that this audience is about to shrink dramatically.
If AI trading agents become the dominant market participants, the product that matters is not a dashboard for human analysts but a data feed and reasoning layer that agents can query programmatically. The firm is rebuilding toward that interface.
The practical shift involves making Nansen’s datasets machine-readable in ways that are optimized for agent consumption, not human visualization.
That means structured APIs, real-time signal streams, and contextual tagging that an agent can ingest and act on without a human intermediary interpreting charts.
Why The Timeline Of 2 Years Is Credible And Why It Matters
Svanevik’s two-year window is aggressive but grounded in observable trends. AI agents capable of executing multi-step onchain actions have already appeared on Ethereum and Solana, and several protocols have built agent-native interfaces designed to let programs interact without wallet pop-up confirmations that require human approval.
The financial stakes of getting this transition right are substantial.
If AI trading agents dominate order flow, they will also dominate the extraction of value from price inefficiencies, a pool that currently generates billions of dollars annually for human traders and trading firms. Firms that position their infrastructure for agent consumption before that transition completes will capture the institutional revenue that follows.
The parallel to the rise of algorithmic trading in equity markets in the 2000s is instructive.
When algorithms came to dominate U.S. equity markets, the firms that survived were those that had retooled their data and execution infrastructure for machine consumption years before the transition was complete. Human traders did not disappear, but their share of total volume shrank permanently.
The Infrastructure Gap AI Agents Must Cross
Several technical obstacles remain before AI trading agents can plausibly dominate crypto markets.
Onchain latency is the most acute: a language-model reasoning step adds milliseconds to hundreds of milliseconds of delay, which matters enormously in liquid markets where arbitrage windows close in under a second.
Key management is a second constraint. An agent that can autonomously move funds must hold private keys, which creates custody risk that most institutions have not resolved.
Hardware security modules and multi-party computation signing schemes are emerging as candidate solutions, but neither is standardized across chains yet.
The regulatory question is also unresolved. If an AI agent executes a trade that a regulator later classifies as market manipulation, liability attribution is murky.
Is the liable party the agent’s operator, the model developer, or the firm that provided the onchain data feed? Frameworks covering autonomous financial actors do not exist in the United States or Europe.
A Rebuilding Race That Has Already Begun
Nansen is not alone in repositioning for an agent-driven market.
Several competing analytics platforms have launched agent-queryable APIs in the past six months. Protocol teams on Ethereum and Solana have begun publishing machine-readable governance summaries specifically designed for agent ingestion.
Svanevik’s public commitment is notable because it sets a hard two-year benchmark against which Nansen’s progress will be measured.
If AI trading agents do not materially dominate market flow by mid-2028, the firm’s product bets will look premature. If the transition happens faster, Nansen’s head start becomes a durable competitive advantage.
The CEO has, in effect, staked the firm’s positioning on a prediction most of his peers have made privately but not yet said aloud.
Read Next: Morgan Stanley’s ETH and SOL ETPs Top Bitcoin’s $381M, the Fee War Starts Here
