Google's delay of Gemini 3.5 Pro erased roughly $200 billion of Alphabet's market value in a single trading session, after the model's coding and complex-reasoning capabilities fell short of internal targets even following a late retraining push with more code-heavy data. The model was unveiled at Google's developer conference in May with a promised June public rollout; as of mid-July it still hasn't shipped, and Google hasn't given a firm new date.
The timing is what makes this more than a single bad product cycle. Anthropic reported run-rate revenue that reportedly reached $47 billion by May 2026, up from roughly $9 billion at the end of 2025 -- growth CEO Dario Amodei has publicly described as "crazy" and "too hard to handle." Whatever the precise accounting nuances (OpenAI has separately disputed parts of Anthropic's revenue methodology), the direction is unambiguous: Anthropic is compounding enterprise revenue at a pace that makes Google's execution stumble look worse by comparison, not just bad in isolation.
“The timing is what makes this more than a single bad product cycle.”
Meanwhile DeepSeek made its 75% price cut to V4-Pro permanent earlier this year, creating a cost gap of 20-35x versus premium Western frontier models for many production workloads. That forces every enterprise AI buyer to re-run procurement math that assumed Western labs would hold pricing power, and it puts pressure on Google, OpenAI and Anthropic alike to justify premium pricing even as DeepSeek's underlying model quality has closed much of the capability gap.
Google is effectively fighting on two fronts simultaneously: a well-capitalized US rival (Anthropic) racing ahead on enterprise revenue and product execution, and a Chinese lab (DeepSeek) racing to the bottom on price while maintaining competitive quality. Neither front rewards a delayed model launch, and Google doesn't have the luxury of choosing which fight to prioritize.
What to watch: whether Google gives a firm new Gemini 3.5 Pro date in its next earnings call, and whether Anthropic's revenue growth rate decelerates from its current "crazy" pace as the law of large numbers eventually catches up.