Best single gauge, not summary. Systematically excluding companies that are large and buzzy yet not profitable is a matter of intentional design to improve the accuracy of the gauge, even if previous companies were not quite this large. Anthropic and OpenAI are great illustrations of why you might want such a design: the bull case for each is that they're going to dunk the other and become the US's primary provider of AI inference, and neither is yet profitable, so by including both of them in the index you're "double counting" investor expectations of how valuable a company producing profitable AI inference will be.