I tested an AI startup that claimed it could beat Wall Street earnings estimates. It won 40 out of 50 times

What happened
In October, the financial AI startup Samaya was challenged to predict third-quarter earnings for 50 major companies before their official results were released. Samaya's AI-generated forecasts were compared to the consensus estimates from Wall Street analysts. The results showed that Samaya's predictions were more accurate than the analysts' in 40 out of the 50 cases. The companies involved represented a range of sectors and included well-known names such as Coca-Cola, Verizon, and Palantir.
Key details
- Samaya raised $43 million from investors in 2025.
- The experiment covered companies reporting Q3 earnings between October 26 and November 5.
- Samaya's predictions used only publicly available data, not private analyst research.
- Samaya exactly matched the actual EPS for Permian Resources and closely predicted revenues for companies like Verizon and Palantir.
- Samaya's clients include banks and hedge funds.
Why it matters
Banks, hedge funds, and other investors who rely on earnings forecasts may now have access to more accurate predictions using AI tools like Samaya. This could reduce their dependence on traditional analyst consensus and potentially change how investment decisions are made. Companies whose stock prices move based on earnings surprises may also see changes in how their results are anticipated and interpreted by the market.
