OpenAI and Anthropic are presenting a version of profitability that excludes their largest expenses: model training and inference. Critics compare this to an airline ignoring the cost of its jets. This financial engineering aims to create a positive outlook for potential IPOs but masks their true cash burn rate.

A significant portion of recent S&P 500 earnings growth is an accounting illusion. In one quarter, Alphabet, Amazon, and Nvidia reported $69B in "non-operating income" simply by marking up their investments in other tech firms. This circular, non-cash gain accounted for a stunning 12% of the S&P 500's total increase.

A portion of Google's cloud revenue and backlog is circular. Google invests in a customer like Anthropic, which then uses that capital to purchase Google's cloud services. This dynamic inflates growth metrics and requires careful scrutiny of revenue quality.

Alphabet holds $122 billion in assets on its balance sheet that are "not yet in service" and thus not yet depreciating. This inflates current Google Cloud margins. Once these assets become operational, depreciation expenses will rise, putting significant downward pressure on future profitability.

Companies buying AI infrastructure capitalize the expense over 5-10 years, while sellers like NVIDIA recognize revenue immediately. This accounting discrepancy creates an illusion of higher S&P profitability, as the same dollar contributes more to profits than it does to expenses on paper.

Anthropic's surprise Q2 profitability could be a strategic maneuver. Evidence from SpaceX's S-1 filing suggests a deal for reduced-fee compute in May and June, perfectly timed to boost financials for an IPO filing and create a favorable but potentially misleading narrative.

It's increasingly difficult to gauge the true profitability of cloud businesses due to circular investments. Tech giants invest in AI startups, which then use that capital (often in the form of cloud credits or vouchers) to pay for compute on the investor's platform, inflating reported revenue growth without a corresponding cash transaction.

Tech giants like Google and Amazon report massive profits partly from paper markups on their investments in AI labs like Anthropic. These labs then spend the investment capital on cloud services from their investors, creating a fragile, self-referential financial ecosystem.

Some tech companies have doubled the depreciable life of their AI hardware (e.g., from 3 to 6 years) for accounting purposes. This inflates reported earnings, but it contradicts the economic reality that rapid innovation is shortening the chips' actual useful life, creating a significant red flag for earnings quality.