# Samaya AI Beats Analyst Estimates at 40 of 50 Companies

By John Kim

Canonical URL: https://www.tokenpost.com/news/business/61206
Published: 2026-11-19T08:28:06.000Z
Updated: 2026-11-19T08:28:06.000Z
Section: Business

> The two-week experiment covered third-quarter earnings from Oct. 26 to Nov. 5 across 10 sectors. Samaya supplied each estimate at least a day before results were released.

Samaya’s forecasts were more accurate than analyst consensus estimates for 40 of 50 companies in an experiment covering third-quarter earnings from Oct. 26 to Nov. 5 across 10 sectors. The comparison used each company’s selected metric, either earnings per share or revenue.

Samaya provided each estimate at least a day before the company released its results. For Verizon, Samaya forecast revenue of $34.52 billion, closer to the actual $34.31 billion than the $34.77 billion analyst consensus. For Palantir, its $2.30 billion revenue estimate was closer to the actual $2.31 billion than the $2.18 billion consensus.

Analysts had more accurate estimates for some companies. Samaya’s $1.42 earnings-per-share estimate for Yum Brands was farther from the actual $1.56 than analysts’ $1.49 estimate. Analysts also did better on estimates for S&P Global and PayPal, and on how far Amazon and Google revenue would fall short of expectations.

Samaya CEO Maithra Raghu said the startup relied on public material, including filings, news reports and interview transcripts. She said the company may add analyst notes to its training data, while cautioning that doing so could teach its system to reproduce analysts’ consensus bias.

Samaya’s clients include banks and hedge funds. The experiment covered one earnings period and 50 companies, leaving open how the results might hold across other reporting periods. Its findings also raise questions about the role of analyst consensus as AI forecasting tools develop.


