Sequoia Partner: Early Underestimation of Anthropic Due to Human Brain's Poor Handling of Exponential Thinking
In a podcast, a partner at Sequoia Capital explained the reasoning behind their $250 million investment in Anthropic, stating that prior judgments must be updated when environmental changes occur, as the human brain struggles with exponential growth, leading to an early underestimation of the company.
He noted that linear thinking is easier to grasp, but predicting exponential changes is challenging. Sequoia had missed opportunities with Anthropic for years, favoring OpenAI and xAI instead, before changing their stance and investing in January 2026.
Sequoia's new co-leads, Alfred Lin and Pat Grady, led the decision-making process, subsequently increasing their bets on the company, making it one of Sequoia's largest investments in history. When asked about specific signals they missed, the partner discussed the process of adjusting their judgments.
This discussion took place on Harry Stebbings' 20VC podcast, focusing on Sequoia's internal reassessment of AI frontier model companies and changes in the competitive landscape.
Driven by events, venture capital is accelerating towards validated exponential growth AI labs, shifting from early caution to large-scale late-stage investments. Beneficiaries include model companies like Anthropic that have demonstrated rapid expansion, while traditional investment frameworks that still evaluate AI based on linear assumptions are under pressure.
Source: Public Information
ABAB AI Insight
Sequoia Capital has long avoided simultaneously betting on direct competitors, previously focusing on supporting OpenAI and xAI while maintaining a cautious stance on Anthropic for years. After Alfred Lin and Pat Grady took over as co-leads at the end of 2025, they adjusted their strategy and participated in Anthropic's financing in January 2026, subsequently increasing their investment in higher valuation rounds.
In terms of capital strategy, Sequoia shifted resources from a single-winner approach to a multi-model parallel layout, motivated by the recognition that the AI foundational model market is large enough to support multiple giants simultaneously, and using later large checks to capture shares during the exponential growth phase. The investment strategy transitioned from early misses to late-stage re-investment, reflecting a repricing of the acceleration in computing and revenue curves.
This is similar to Sequoia's early missed investment in Facebook followed by later investments, or late-stage top-ups for exponential companies like Google; currently, we are in a capital-intensive period for AI, where the valuations and financing scales of leading model companies far exceed traditional software paradigms.
Essentially, this represents capital concentration. Exponential growth forces investment institutions to abandon linear priors, as the mechanism dictates that when the technology curve is steep, the cost of early underestimation is very high, necessitating larger amounts of capital later to compensate for cognitive lag, thereby further concentrating capital on a few validated exponential players.
ABAB News · Law of Cognition
Linear brains always underestimate exponential
Outdated priors must be updated immediately
If you miss early, make up with scale