Karp and Zitron warn AI boom is a bubble: $50B sustainable revenue

Recently, the AI industry saw contrasting visions for its future. The first, epitomized by SpaceXAI's IPO, imagines trillion-dollar frontier companies and data centers in space. But a more grounded scenario has emerged from two CNBC guests: Palantir CEO Alex Karp and analyst Ed Zitron. Karp criticized the centralized model where corporate customers host models with OpenAI and Anthropic. These providers can see into customers' businesses and clone successful use cases. Palantir positions itself as a secure alternative, hosting open-weight models with the same compartmentalization it offers government clients.

Zitron agreed with Karp's criticism and added that the AI industry's economics are broken. OpenAI and Anthropic's current revenue cannot finance the massive costs of training future models. He expects large-scale cancellations of data center construction as demand fails to materialize. A sustainable AI industry, he argues, might generate less than $50 billion per year. The article's author finds this Karp-Zitron scenario more plausible than the trillion-dollar hype, especially as governments may shroud advanced capabilities for national security, leaving a publicly commercialized sector that resembles China's more decentralized AI landscape.