Major customers of frontier AI labs, such as voice AI company Eleven Labs, are actively working on proprietary models. This trend of verticalized model development signals a desire to escape data leakage concerns and dependence on potential future competitors.

The assumption that enterprises need the most advanced 'frontier' models is being challenged. For many common business tasks like email analysis or processing limited data sets, less powerful and more cost-effective models are sufficient. This trend poses a direct threat to the premium pricing strategy of companies like Anthropic.

Recent data from Ramp shows frontier models' usage share fell from 53% to 45% in a single month, while standard models gained share. This indicates a market shift towards cost-effectiveness and "good enough" performance over cutting-edge capabilities for many use cases, challenging the moat and pricing power of companies like OpenAI and Anthropic.

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.

As part of its 'token minimizing' strategy, Meta is encouraging employees to use its in-house tools like MetaCode over more advanced external models. This creates an awkward trade-off: potentially reducing employee productivity to lower the company's massive AI operational expenditure bill.

Despite being a major cloud partner, Microsoft is actively developing its own frontier AI models to compete with and reduce dependency on third-party labs. AI chief Mustafa Suleiman called Anthropic's models "extremely expensive" and stated the company's goal is to eliminate this cost.

Following an unmonitored spending surge during early AI adoption, Meta and Microsoft are reining in employee use of Anthropic's Claude. Microsoft, which was on track to spend $1 billion annually on Claude, reduced spending by over 33% by redirecting engineers toward OpenAI models and internal models. First-party and contracted models eliminate third-party software licensing markups, leaving companies responsible only for the raw electricity and compute required for inference.

As companies build successful products like Muse on cheaper, non-frontier models, the revenue growth for premium providers like OpenAI weakens. This trend indicates that the market for standard models is highly competitive, posing a threat to frontier labs' IPO valuations.

Concerns over profit margins are pushing businesses to explore cost-effective AI. This includes using smaller models from giants like OpenAI and Anthropic (e.g., GPT-mini, Haiku), open-source options, or developing in-house models, rather than exclusively relying on the most powerful, expensive versions.