AI-related debt issuance around 800 trillion won... Claude, leading model companies score 2 points

Anthropic’s Claude rated the predictability of leading AI model developers OpenAI and Anthropic at just 2 out of 10, with Oaktree Capital co-founder Howard Marks saying in a memo published on the 11th that this means the businesses are effectively impossible to forecast. Claude scored the predictability of AI as a whole at about 3 out of 10, using a scale from 1 (unknowable) to 10 (fully predictable) that measures how well future business can be gauged rather than investment grade or loss probability. The assessment matters because it underscores how little visibility investors have into the revenue trajectories and competitive positioning of the largest private AI developers as capital spending and debt issuance tied to AI continue to scale. What to watch next is whether forthcoming disclosures from OpenAI and Anthropic, and the pace of enterprise adoption, give investors enough data to raise those predictability scores.