Google and Amazon's AI earnings make the Frontier Case - frontier access is the actual separator

Stratechery's Ben Thompson analyzes Q2 2026 earnings from Google and Amazon. His "Frontier Case": frontier model access creates compounding advantage that commodity AI cannot replicate. The earnings support it.

In plain terms: Google and Amazon reported Q2 earnings. The key insight from Stratechery: the AI companies actually winning aren't the ones adding AI features - they're the ones with access to frontier models. That access is becoming a business moat.

The fact

Stratechery's Ben Thompson analyzed Q2 2026 results from Google and Amazon, centering his analysis on what he calls "the Frontier Case" - the argument that access to the most capable AI models creates compounding competitive advantage over organizations using commodity AI. Both companies reported AI-driven acceleration in their cloud divisions, with Google Cloud and AWS each citing frontier LLM adoption as a primary revenue driver.

Our take

The Frontier Case is a direct answer to the "AI is commoditizing" narrative. Thompson's framing argues that frontier model access (Claude, Gemini, GPT-4o-class systems) doesn't just add features - it creates lock-in through capability compounding rather than switching costs. If this is correct, the race to secure OpenAI, Anthropic, and Google partnerships isn't about features or cost optimization: it's about positioning for a winner-takes-most dynamic in enterprise AI infrastructure. The earnings data from Google Cloud and AWS suggest this dynamic is already playing out in the numbers, not just in pitch decks.

What to watch

Microsoft's Copilot revenue disclosure in its Q2 results - if frontier-vs-commodity divergence shows up there too, the thesis hardens considerably.