Goldman Sachs’ tech boss says tracking individual AI usage isn’t useful. He just watches how fast his 12,000 engineers move from idea to production

Goldman Sachs' technology leadership reveals the firm built an internal ChatGPT alternative and measures AI adoption success through engineering velocity rather than individual usage tracking. The approach prioritizes outcomes—how quickly ideas reach production—over granular user metrics, reflecting a broader enterprise shift toward outcome-based AI governance.

Goldman Sachs' decision to build proprietary AI infrastructure and adopt velocity-based performance metrics signals institutional confidence in AI's strategic importance while challenging conventional monitoring approaches. Rather than tracking individual ChatGPT interactions across 12,000 engineers, the firm focuses on end-to-end delivery cycles, suggesting that aggregate productivity gains matter more than usage surveillance. This methodology reflects mature organizational thinking: when adoption is already widespread, granular tracking becomes noise rather than insight.

This approach emerges from the broader enterprise AI wave where financial institutions compete for technical talent and operational efficiency. Building in-house AI systems allows Goldman Sachs to maintain proprietary advantages, reduce dependency on third-party vendors like OpenAI, and customize models for financial modeling and risk analysis—domains where competitive edge directly impacts profitability. The emphasis on velocity aligns with DevOps culture that dominated software engineering over the past decade, now extending to AI workflows.

For the broader market, this signals that enterprise AI adoption has moved beyond pilot phases into normalized operations. When major financial institutions measure AI success through engineering speed rather than curiosity metrics, it indicates mainstream acceptance. This validates AI infrastructure spending and suggests sustained demand for compute, cloud services, and specialized AI development platforms.

Investors should monitor whether other major institutions adopt similar velocity-based frameworks, as this could accelerate AI tool consolidation and increase competition among AI platform providers seeking enterprise integration. The success of Goldman Sachs' internal model will likely determine whether other firms build proprietary systems or rely on external APIs.

• →Goldman Sachs built proprietary AI infrastructure rather than relying solely on third-party solutions like OpenAI
• →Measuring engineering velocity and idea-to-production speed proves more valuable than tracking individual AI tool usage
• →Enterprise AI adoption has transitioned from experimental pilots to core operational workflows across large technical teams
• →In-house AI systems provide financial institutions with competitive advantages in proprietary modeling and risk analysis
• →Velocity-based metrics suggest mature organizational AI integration focused on tangible business outcomes rather than adoption vanity metrics