OpenAI
OpenAI demonstrates that Pacific Salmon strategy - grow at all costs, die trying - is correct in exactly 20% of situations.
OpenAI demonstrates that Pacific Salmon strategy - grow at all costs, die trying - is correct in exactly 20% of situations. During the AI boom (2023-present), winner-takes-all dynamics and closing market windows justify burning capital to achieve dominance before the power law locks in.
Before ChatGPT's November 2022 launch, OpenAI maintained 25-30% annual turnover, deliberately importing talent from DeepMind, Google Brain, academia, robotics, and gaming. This high migration rate accelerated exploration of transformer architectures and RLHF by bringing diverse ML approaches into collision. Once ChatGPT achieved product-market fit, turnover dropped to 12-15% to stabilize the scaling phase.
The lesson: optimal gene flow varies by lifecycle stage. During exploration, high turnover imports diverse approaches and accelerates innovation. During exploitation, reduced turnover preserves accumulated knowledge and enables scaling. OpenAI understood that the rules change when you shift from searching for product-market fit to defending it.
OpenAI Appears in 2 Chapters
OpenAI exemplifies the 20% of situations where Pacific Salmon growth-at-all-costs strategy is appropriate due to winner-takes-all dynamics.
When OpenAI's burn rate strategy makes sense →OpenAI maintained 25-30% turnover during exploration (2019-2022), then reduced to 12-15% post-ChatGPT to stabilize and scale.
How OpenAI modulated talent migration by lifecycle stage →