Scaling Limits Emerge

The second-most likely outcome is that Q Labs’ result is real but does not generalize cleanly to larger models, richer data, or standard evaluation suites. This would preserve the relevance of backpropagation and Adam-style optimizers in frontier AI training.

Each figure is the modeled move of the S&P 500 if the outcome occurs. For information only. Not investment advice.