Wall Street is starting to pay attention to one of crypto’s most ambitious AI plays.
For months the big question in AI investing has been deceptively simple: if large language models can replace lawyers, consultants and coders, can they replace hedge fund managers? That debate is moving out of theory and into live markets, and at the center of the action is Nof1 — an AI trading lab backed by Nasdaq-listed SUI Group (SUIG) and London prop shop Karatage.
Nof1’s flagship experiment, Alpha Arena, puts frontier models from the likes of OpenAI, Anthropic, Google and xAI into head-to-head autonomous trading competitions using real capital. The early results have been messy: most models lost money, overtraded or failed to manage risk. For sophisticated investors, though, that failure is precisely the signal they want to see. It exposes the gap between language understanding and the unique demands of financial markets — timing, position sizing, adaptability and robust risk control.
The easy AI trade over the past two years has been to back companies building chatbots and copilots. The harder, and potentially much bigger, opportunity lies in AI systems that can operate in adversarial, fast-moving environments like trading, where prediction alone doesn’t cut it. Trading rewards consistent capital allocation under uncertainty, not just fluent language — making it a tougher next frontier for AI.
Nof1’s thesis is straightforward: today’s frontier models were trained to predict language, not to allocate capital. Alpha Arena aims to close that gap by exposing models to live market outcomes that reward successful behavior and punish mistakes. That hands-on training approach helps explain why SUI Group and Karatage co-led Nof1’s $15 million funding round. The pair have also backed Recursive Superintelligence alongside TwinPath Ventures — a self-improving AI startup reportedly valued above $4 billion — indicating a coordinated thesis on where AI and markets are converging.
Both Nof1 and Recursive draw from open-endedness research, a branch of AI work that focuses on systems that can continually innovate and improve rather than converge on a fixed task. The bet is simple: if autonomous agents become central to investing, the teams building the infrastructure and live training environments now will help shape the future of finance.
After season two of Alpha Arena, Nof1 plans to launch a consumer product it bills as the world’s first “coding agents for markets.” The platform will surface the lab’s frontier models — developed with open-ended AI architectures designed to keep improving — with the stated aim of exceeding both human and algorithmic traders. For crypto-native investors and institutions looking to marry blockchain infrastructure with next-gen AI, Nof1 represents a high-risk, high-reward experiment that Wall Street is no longer willing to ignore.
Read more AI-generated news on: undefined/news
