According to @balajis, the widely shared quartile chart suggesting many AI companies can reach 100 million ARR in about three and a half years is an extrapolation from growth rate buckets rather than an empirical cohort of roughly one thousand startups, which changes how investors should interpret the distribution of outcomes, source: @balajis on X, Sep 23, 2025. According to @balajis, a true cohort that includes the many AI startups that never reach one million ARR would show that achieving 100 million ARR in three and a half years remains exceptional and is likely in the top five to top one percent depending on how an AI startup is defined, source: @balajis on X, Sep 23, 2025. According to @balajis, he points to Midjourney and Cursor as examples on this trajectory and notes that Replit and OpenAI would not fit the three year window because they were founded earlier, source: @balajis on X, Sep 23, 2025. According to @balajis, reaching 72 million ARR in five years with strongly positive margins is also a phenomenal outcome and should not be treated as ordinary, source: @balajis on X, Sep 23, 2025. According to @balajis, traders in AI themed equities and tokens should treat extrapolated quartiles as aggressive scenarios rather than base cases, adjust screening and valuation frameworks for survivorship bias, and avoid assuming a 100 million in three and a half years ramp without high outlier probabilities, source: @balajis on X, Sep 23, 2025.