I'm not convinced the AI bubble exists

The cost to access and use leading AI models may not increase as rapidly as some anticipate.

The author argues against the prevailing notion of an "AI bubble," asserting that the substantial investments in AI companies are justifiable and that further growth is likely. While acknowledging the high expenses, particularly for data center build-outs, the article posits these are largely one-off capital expenditures. Once infrastructure is in place, the operational costs of AI are presented as profitable, citing Anthropic's revenue-to-compute cost ratio. The author suggests that companies can manage significant debt obligations, estimated at $5 trillion over 20 years, by leveraging substantial revenues, pointing to Alphabet's earnings as a benchmark. Rapid revenue growth, as demonstrated by Anthropic, suggests strong market demand that is far from saturated. The argument also dismisses the "no moat" concern, comparing AI's competitive landscape to the chicken sandwich wars, where profitability was achievable despite easy product replication. Finally, the author contends that AI companies will continue to attract capital through venture funding, public debt markets, and eventually, profits, indicating a sustainable growth trajectory rather than an impending collapse.

The assertion that AI companies are spending unsustainably is challenged by framing data center costs as a one-time capital expenditure, rather than ongoing operational expenses. Once this infrastructure is built, the article suggests that AI's per-unit cost of operation is highly profitable, potentially allowing companies to manage significant debt loads. This perspective implies that current high valuations may be grounded in future operational efficiency and profitability rather than speculative growth.

The article refutes the idea that a lack of competitive "moats" for AI companies will lead to their downfall. It argues that brand loyalty and institutional lock-in, rather than proprietary technology, will allow companies to maintain market share and generate revenue. This suggests that the competitive landscape, while fierce, may still support a few dominant players who can profit significantly from widespread AI adoption across various industries and consumer uses.