‘Selling coffee beans to Starbucks’ – how the AI boom could leave AI’s biggest companies behind

The competitive landscape of the AI industry is undergoing a significant shift, with startups increasingly commoditizing foundation models (FMs) due to diminishing returns from pre-training scale, opting instead to focus on application-layer fine-tuning and specialized solutions. This trend threatens to relegate major FM developers like OpenAI and Anthropic to low-margin backend suppliers, as the perceived technological moat for FMs diminishes and first-mover advantages prove fleeting. Consequently, large investments in foundational models, such as Meta's, are becoming riskier, despite some FMs retaining advantages in brand recognition and infrastructure.

Analysis

The competitive landscape in artificial intelligence is undergoing a structural shift, as the strategic advantage once held by foundation model (FM) developers like Google (GOOG/GOOGL) and OpenAI is diminishing. The primary driver is the observation of diminishing returns from the scaling of pre-training, which is redirecting focus towards application-layer innovations such as fine-tuning and interface design. This trend is leading to the commoditization of FMs, which are increasingly viewed by startups as interchangeable backend components, thereby eroding the pricing power and potential moat of their creators. Evidence suggests a weak first-mover advantage, with venture capitalists noting that early leads in specific AI categories have been quickly lost to competitors. This dynamic poses a significant risk to the capital-intensive strategies of major technology firms, particularly Meta (META), whose multi-billion dollar AI spending is described as increasingly risky, a sentiment reflected in its negative ticker score (-0.6). While large FM developers retain advantages in brand, infrastructure, and capital, the current trajectory suggests their role could be relegated to that of a low-margin supplier, fundamentally altering the investment thesis that has driven high valuations in the sector.

Key Decisions for Investors

• Investors should critically re-evaluate the long-term margin potential for companies making massive capital expenditures on proprietary foundation models, such as Meta and Google, given the increasing risk of commoditization.
• Consider increasing exposure to application-layer AI companies or 'AI wrappers' that demonstrate strong product-market fit and are model-agnostic, as value appears to be migrating from the core infrastructure to these specialized solutions.
• Monitor the pricing power of leading foundation models and the adoption rate of open-source alternatives, as these are key indicators of whether a defensible moat exists at the AI infrastructure level or if value will continue to accrue at the application layer.