速递｜从LLM到LTM：Fundamental以“数据基础模型”切入，A轮融资2.55亿美元

Core Insights - Fundamental, an AI lab, has launched a new foundational model called Nexus aimed at extracting insights from structured data generated by enterprises, addressing a long-standing issue in data analysis [2] - The company has raised $255 million in funding, concluding its stealth mode with a valuation of $1.2 billion, attracting significant investor interest [3] Group 1: Model Features and Innovations - Nexus is referred to as a large table model (LTM) rather than a large language model (LLM), and it offers determinism, providing the same answer for the same question every time [3] - Unlike most contemporary AI models based on the Transformer architecture, Nexus does not rely on this framework, which allows it to handle large-scale datasets more effectively [3][4] Group 2: Market Opportunities and Applications - The model addresses the limitations of Transformer-based AI models, which struggle with reasoning over large datasets, such as spreadsheets with billions of rows, creating significant opportunities for enterprises [4] - Fundamental's Nexus allows for the application of modern technology to big data analysis, promising superior performance compared to traditional algorithms and enabling scalability across various use cases [4] Group 3: Strategic Partnerships and Contracts - The company has secured high-value contracts, including seven-figure agreements with Fortune 100 companies, indicating strong market demand for its solutions [4] - Fundamental has established a strategic partnership with AWS, allowing AWS users to deploy Nexus directly from existing instances [4]