This tension is mounting as Google Cloud sells significant TPU capacity to external competitors, including the AI startup Anthropic.
The internal friction highlights a strategic conflict between Google’s research goals and its rapidly expanding cloud business, which recently reported 82% revenue growth.
While the cloud division profits by providing infrastructure to third-party AI labs, internal teams feel sidelined by the commercialization of the very hardware they helped pioneer.
This atmosphere of scarcity and bureaucracy has contributed to a high-profile talent exodus; notably, all eight authors of the foundational 2017 "transformer" research paper have now left the company, alongside veteran chief scientist Jeff Dean, who departed after 27 years to launch a new startup.
To address these shifting dynamics, Google has reorganized its leadership, moving DeepMind co-founder Demis Hassabis to a chairman and chief scientist role while Koray Kavukcuoglu takes over daily management of AI model development.
The company is now attempting to balance the massive capital expenditure—the funds used to acquire or upgrade physical assets like data centers—required for cutting-edge research against the immediate profitability of its cloud services.
This transition affects the entire AI industry, as Google’s decisions on whether to prioritize its own Gemini models or support external partners like Anthropic will dictate the pace of innovation across the sector.