Google published research on automating scientific paper review using AI assistance tools. The work applies language models to structured aspects of peer review workflows—likely abstract summarization, consistency checking, and preliminary technical assessment. For research infrastructure builders, this signals that peer review—historically resistant to automation due to its epistemic and social complexity—is now being systematically decomposed into automatable subtasks. This reduces the friction for platforms attempting to accelerate review cycles or scale review capacity without proportional reviewer recruitment. Operationally, builders of scientific AI tools face a narrowing window: institutions will soon integrate automated pre-review filtering and assessment aids. This compresses margins for standalone peer review assistance products while creating demand for integration with existing review management systems. The second-order effect is structural—institutions gain leverage to enforce stricter submission standards and faster rejection pipelines, shifting cost from human reviewer time to infrastructure cost. For teams building scientific platforms, assuming peer review remains a major operational bottleneck may no longer hold.