AI infrastructure leaders justify massive investments by citing a limitless appetite for intelligence, dismissing concerns about efficiency. This belief ignores that infinite demand doesn't guarantee profit; it can easily lead to margin collapse and commoditization, much like the internet's effect on media.

Creating frontier AI models is incredibly expensive, yet their value depreciates rapidly as they are quickly copied or replicated by lower-cost open-source alternatives. This forces model providers to evolve into more defensible application companies to survive.

Contrary to the AI growth narrative, immense CapEx is transforming 'cap-light' tech giants into capital-intensive businesses. This spending pressures margins, reduces returns on capital, and mirrors historical capital cycles where infrastructure builders rarely reaped the primary rewards.

Similar to how blockchain protocols like Bitcoin and Ethereum accrued more value than the apps built on them, AI foundation models are getting 'fatter.' They are absorbing more capabilities, allowing users to perform complex tasks in a single step within the base model, reducing the need for specialized application-layer companies.

The massive, redundant CapEx in AI infrastructure is analogous to the late-90s fiber-optic boom. While that fiber enabled future giants like Netflix, the initial investors went bankrupt. This suggests the ultimate beneficiaries of AI may be society and end-users, not the companies spending trillions on the build-out.

If AI makes intelligence cheap and universally available, its economic value may collapse. This theory suggests that selling raw AI models could become a low-margin, utility-like business. Profitability will depend on building moats through specialized applications or regulatory capture, not on selling base intelligence.

Unlike cable or power companies that benefit from regional monopolies, AI intelligence is a globally competitive, frictionless market. This dynamic is 'so much worse' for business because it allows for perfect arbitrage, driving the price of intelligence toward zero and making it incredibly difficult to build a sustainable, high-margin business on the infrastructure layer.

Despite billions in funding, large AI models face a difficult path to profitability. The immense training cost is undercut by competitors creating similar models for a fraction of the price and, more critically, the ability for others to reverse-engineer and extract the weights from existing models, eroding any competitive moat.

Sam Altman's analogy of selling AI like electricity is flawed because utility providers are low-margin businesses. Without strong differentiation, model labs will face price competition, becoming a commodity. The real value will be captured by applications built on top, just as apps, not telcos, captured mobile's value.