Contrary to the narrative of burning cash, major AI labs are likely highly profitable on the marginal cost of inference. Their massive reported losses stem from huge capital expenditures on training runs and R&D. This financial structure is more akin to an industrial manufacturer than a traditional software company, with high upfront costs and profitable unit economics.

While increased CapEx signals strength for cloud providers like Microsoft and Google (who sell that capacity to others), the market treats Meta's spending as a pure cost center. Every dollar Meta spends on AI only sees a return if it improves its own products, lacking the direct revenue potential of a cloud platform.

A contrarian view argues that encouraging high token usage ("token maxing") is a valid short-term strategy. The rationale is that the engineering challenge of building systems capable of consuming tokens at massive scale is a significant achievement and a proxy for deep AI integration, making the raw cost secondary.

Meta's huge AI capex, despite no hit product yet, is based on proprietary data from its massive platform. Unlike the speculative Metaverse venture, this investment is a direct response to observed exponential growth in user engagement with AI content, even if users publicly claim to dislike it.

Initial estimates placed Meta's monthly Anthropic bill near a billion dollars. However, a breakdown reveals that since most tokens are low-cost inputs (code context) rather than high-cost outputs, the actual monthly cost is likely between $55M and $136M—substantial, but a fraction of the headline figure.

Meta's massive internal consumption of AI tokens for tasks like code generation creates a multi-billion dollar expense. By developing its own frontier models in-house, Meta can vertically integrate, justifying the high cost of its AI lab (MSL) purely on internal savings, even before launching any new consumer AI products.

Critics argue AI revenue must grow exponentially to justify investment. However, for incumbents like Meta, this isn't net-new revenue. It's a massive internal budget shift from established products to new AI features, redirecting existing user engagement and spend rather than creating a market from scratch.

In the AI era, token consumption is the new R&D burn rate. Like Uber spending on subsidies, startups should aggressively spend on powerful models to accelerate development, viewing it as a competitive advantage rather than a cost to be minimized.

Meta's multi-billion dollar super intelligence lab is struggling, with its open-source strategy deemed a failure due to high costs. The company's success now hinges on integrating "good enough" AI into products like smart glasses, rather than competing to build the absolute best model.