Early Trends and PMF in AI-for-Hedge Funds Startups (magis.substack.com)

Ken Griffin’s recent public skepticism—that generative AI boosts productivity but “falls short” of reliably producing investment alpha—has tempered hype around the surge of AI-for-hedge-fund startups, even as the field attracts heavy interest from buy-side engineers and Silicon Valley founders. I track 100+ private companies in the space, and most lack visible customers or revenue and haven’t reached clear product–market fit. A standout exception is AlphaSense, which predates the ChatGPT era, has expanded via acquisitions (including Tegus in 2024), added an AI research copilot and automated spreadsheet analysis, and reported >$500M ARR in Oct 2025—though exactly how much growth is AI-driven remains unclear. Technically, the market is coalescing into distinct approaches: research copilots that answer free-form investment questions and automate due diligence; Excel copilots that generate or update financial models from natural-language prompts; “Terminal 2.0” interfaces that layer LLM-driven summaries, alerts and context over real-time data; AI model providers building time-series/quant models and signal APIs for quants; and data-extraction tools that turn SEC filings, transcripts, and PDFs into structured datasets. These categories blur in practice, and PMF may ultimately come from middle/back-office automation, proprietary forecasting models, or highly verticalized data pipelines rather than from general-purpose alpha-generating LLMs.