Shopify CTO: 100% AI Adoption, 30% Monthly Merge Growth, PR Review Now the Bottleneck
Shopify CTO Mikhail Parakhin appeared on the Latent Space podcast to reveal concrete AI adoption metrics: 100% of Shopify's engineering organization now uses AI, with 30% month-on-month growth in code merges. Unlimited token budgets for all engineers have been in place throughout, with usage slope changing sharply in December 2025. The current constraint is no longer code generation — it's PR review and CI/CD throughput. Shopify has built SimGym (a customer simulation system using Shopify's historical transaction scale) and two new internal experimentation tools, Tangle and Tangent, to manage the volume of AI-generated changes. Larger models are proving cheaper in the long run, with unlimited tokens and auto-research loops unlocking gains across every engineering domain.
Why It Matters
Shopify's data is one of the most concrete enterprise-scale AI engineering datasets publicly shared. The "PR review is the new bottleneck" finding directly challenges the assumption that code generation is where AI impact concentrates — and suggests that investing in review tooling, CI/CD throughput, and test infrastructure is the next highest-leverage area for AI-accelerated engineering teams. Episode: Latent Space.