Milk Man and lead researcher Martin examine the bear case for the AI bottleneck trade, centered on whether OpenAI and Anthropic can generate enough revenue to justify massive compute commitments. One estimate cited in the discussion says the two frontier labs need to reach at least $180 billion in combined revenue by the end of 2026 for the trade to continue as expected.
The risk is that enterprises and consumers increasingly choose cheaper open-source models, internal AI systems, specialized tools, and agents distributed by companies such as Meta and SpaceX. If frontier-lab growth slows, excess compute supply could pressure cloud providers, hyperscalers, and other infrastructure companies tied to the buildout.
Martin remains bullish on AI overall but argues that adoption does not guarantee gains across every part of the market. Lower compute costs could instead favor application-layer companies already using AI to improve products, cut costs, and expand margins.
VIDEOS
Join 330,000+ investors finding today’s opportunity for tomorrow’s wealth.
Get started today and sign up to any of our three newsletters
