NewsCryptoNansen CEO Predicts AI Trading Agents Will Outnumber Human Traders Within Two Years

Nansen CEO Predicts AI Trading Agents Will Outnumber Human Traders Within Two Years

Author: Coinotag·

Key Takeaways

  • Nansen CEO Alex Svanevik forecasts that AI trading agents will surpass human traders in number within roughly two years, referring to active software agents rather than superior investment returns.
  • Nansen's current trading product requires human approval before execution, with AI handling research and order preparation while users retain final review of price, size, route, and fees.
  • An experimental autonomous agent generated only $23 in profit while consuming $700 in large language model inference costs, demonstrating that fully independent AI trading is not yet commercially viable.
  • Approximately two-thirds of the fifteen most actively traded perpetual contracts on Nansen track non-crypto reference assets such as SpaceX private equity, the S&P 500, gold, and crude oil rather than digital tokens.
  • Nansen has integrated Robinhood into its platform with trading functionality expected to go live within days, though the announcement did not specify custody arrangements, routing details, or regional eligibility.
Nansen CEO Predicts AI Trading Agents Will Outnumber Human Traders Within Two Years

Bitcoin (BTC) stands as the primary market-structure asset affected by a forecast from Nansen co-founder and chief executive Alex Svanevik, who anticipates that AI trading agents will outnumber human traders within approximately two years. The prediction pertains to the number of active software agents rather than any assertion that AI will generate superior returns. The distinction matters because crypto markets already host substantial automated flow from MEV extractors, arbitrage bots, and grid-trading programs that execute on deterministic rules; the newer category Svanevik describes involves large language model-based agents capable of interpreting unstructured data such as social sentiment, news, and on-chain activity before deciding whether to act.

Svanevik noted that Nansen's trading function has processed over $500 million in cumulative volume since launch, though that figure reflects repeated buy and sell orders rather than net profit or assets under management. The company's current product maintains a human-in-the-loop design: an AI trading bot can research tokens, trace large wallets, and prepare order parameters, but the user must review price, size, route, fees, and slippage before execution proceeds. Official product documentation describes spot trading on Base and Solana, while perpetual contracts are routed through Hyperliquid. The interface employs an embedded AI crypto wallet powered by Privy rather than allowing direct import of an external wallet.

Svanevik referenced an experimental agent that earned $23 while consuming $700 in inference cost, demonstrating that autonomous execution remains economically fragile. Inference cost refers to the per-query expense of calling large language model APIs, which accumulates each time an agent reads data, reasons over it, and generates a decision. Because crypto markets operate continuously, a single firm or user can deploy separate agents for monitoring, news parsing, risk limits, and execution, meaning agent counts can rise faster than the number of human account holders. This distinction carries implications for market quality, as increased automated order flow does not automatically translate into better price discovery or lower drawdowns.

Nansen has framed fully autonomous agents as subjects of back-testing and simulated trading rather than vehicles for live customer capital. The company's risk notes cite incorrect model output, contaminated on-chain or social data, leverage-driven liquidations, smart-contract failure, and irreversible blockchain settlements, with final responsibility resting on the user. These risks echo long-standing concerns in traditional algorithmic trading—flash crash cascades and feedback loops—while adding crypto-specific vulnerabilities such as irreversibility of settled transactions and dependency on third-party model providers.

A second development involves Nansen's effort to transform analytics into an execution layer for Bitcoin (BTC), Ether, and non-crypto reference assets. Svanevik outlined three strategic shifts: transitioning from research to trade execution, expanding from digital assets to all asset classes, and enabling AI agents to make decisions previously handled by human investors.

Svanevik contrasted human retail behavior with machine diversity, arguing that retail traders frequently chase the same themes while agents can simultaneously process different data sources and models. He acknowledged that large language models can be misled by poisoned inputs, making data security and model reliability prerequisites for wider release.

According to the company's product data, approximately two-thirds of the 15 most actively traded perpetual contracts were not crypto instruments but reference assets tied to SpaceX private equity, the S&P 500, gold, silver, WTI crude, and Brent crude. These are on-chain derivatives tracking external prices; they do not grant holders direct ownership of shares, index constituents, or physical commodities. Such instruments represent a segment of the broader real-world-asset and synthetic-derivatives landscape, where platforms including Synthetix and Polymarket have offered on-chain price exposure to non-crypto references.

Svanevik also confirmed that Robinhood has been integrated into Nansen this week, with trading functionality expected to go live within days, though the announcement did not specify custody arrangements, routing details, or regional eligibility. Robinhood, which serves millions of retail brokerage accounts, has been gradually expanding its crypto offerings, and the integration signals Nansen's push to bridge on-chain analytics with mainstream brokerage rails.

The executive argued that centralized exchanges may adapt more slowly because licensing requirements, supervision obligations, and legacy revenue lines constrain the deployment of autonomous trading tools. Nansen's claimed competitive advantage is its labeled-address database, developed over six years and covering more than 500 million blockchain addresses, which can assist an altcoin trader or macro desk in identifying wallet clusters invisible to ordinary chart readers.

The company continues to test fully autonomous agents through back-tests and simulated trades before committing customer funds.

According to COINOTAG data, BTC dominance stands at 69.8% with the total crypto market capitalization near $1.85 trillion, while the Fear and Greed Index at 29 out of 100 signals fear rather than euphoria. In that context, automated execution could deepen liquidity during calm conditions but potentially amplify one-sided positioning during market shocks.

The disclosed $23 gain against $700 in inference cost indicates that autonomy is not yet commercially mature. Until agents demonstrate robustness after accounting for fees, slippage, and data-poisoning attempts, Bitcoin's macro sensitivity rather than machine count will drive price movements toward or away from all-time highs.

COINOTAG does not provide financial advisory services. This content is for informational purposes only and should not be considered investment advice. Cryptocurrency investments involve high risk.