AI startup lab Fundamental has officially exited its stealth mode. The company announced it has successfully raised $255 million in Series A funding, with a post-money valuation of $1.2 billion. This round was led by Oak HC/FT, Valor Equity Partners, Battery Ventures, and Salesforce Ventures. Several industry giants, including Perplexity CEO Aravind Srinivas and the CEO of Datadog, also participated in the investment personally.

Fundamental's core product is a foundational model called Nexus. Unlike mainstream large language models (LLMs) such as ChatGPT or Claude, Nexus is defined as a "large table model" (LTM). CEO Jeremy Fraenkel pointed out that while existing LLMs perform exceptionally well in handling unstructured data such as text, audio, and video, they often struggle when dealing with the vast and complex structured data (such as various data tables) within enterprises.

In terms of technical architecture, Nexus does not use the Transformer architecture that has driven the current AI boom but instead is a deterministic model. This means that for the same question, it will always provide the exact same answer, which is crucial for rigorous data analysis scenarios. As it is not limited by the context window of traditional architectures, Nexus can easily handle ultra-large datasets containing billions of rows of data, solving the logical confusion problems that traditional AI often faces when dealing with massive tables.

Currently, Fundamental has shown strong commercial monetization capabilities, having signed seven-figure contracts with several Fortune 100 companies. In addition, the company has formed a strategic partnership with AWS, allowing users to deploy the Nexus model directly on their existing cloud instances. Fraenkel stated that the goal of Nexus is to replace large teams of data scientists, enabling companies to efficiently handle diverse analytical use cases through a single model.

Key points:

  • 💰 Valuation enters the unicorn club: With $255 million in Series A funding, Fundamental achieved a $1.2 billion valuation upon exiting its stealth mode.

  • 📊 Focused on structured data analysis: Launching the large table model (LTM) Nexus, which addresses the performance gaps of traditional large language models when handling extremely large data tables.

  • 🛠️ Abandoning traditional architecture to achieve determinism: Not relying on the Transformer architecture, it ensures the accuracy and rigor of data analysis through deterministic output, supporting the processing of massive data sets with billions of rows.