An AI lab called Fundamental emerged from stealth on Thursday, offering a new foundation model to solve an old problem: how to draw insights from the huge quantities of structured data produced by enterprises. By combining the established systems of predictive AI with more contemporary tools, the company believes it can reshape how large enterprises analyze their information.
CEO Jeremy Fraenkel explained that while large language models have excelled at working with unstructured data like text, audio, video, and code, they struggle with structured data such as tables. He stated that with their model, called Nexus, they have built the best foundation model to handle that specific type of data.
This idea has attracted significant investor interest. The company is launching with $255 million in funding and a $1.2 billion valuation. The majority comes from a recent $225 million Series A round led by Oak HC/FT, Valor Equity Partners, Battery Ventures, and Salesforce Ventures. Hetz Ventures also participated, alongside angel funding from Perplexity CEO Aravind Srinivas, Brex co-founder Henrique Dubugras, and Datadog CEO Olivier Pomel.
Fundamental’s Nexus is described as a large tabular model rather than a large language model. It breaks from contemporary AI practices in key ways. The model is deterministic, meaning it provides the same answer every time it is asked a given question. It also does not rely on the transformer architecture that defines models from most modern AI labs. The company calls it a foundation model because it undergoes pre-training and fine-tuning, but the result is profoundly different from what a client would get from partners like OpenAI or Anthropic.
These differences are crucial for the use case Fundamental is targeting, where current AI models often struggle. Transformer-based models can only process data within their context window, making it difficult for them to reason over extremely large datasets, such as a spreadsheet with billions of rows. This kind of enormous structured dataset is common within large enterprises, creating a significant opportunity for models that can operate at that scale.
Fraenkel sees this as a major opportunity. Using Nexus, the company can apply contemporary techniques to big data analysis, offering something more powerful and flexible than the algorithms currently in use. He explained that clients can now have one model across all their use cases, massively expanding the number of problems they can tackle. On each use case, they can achieve better performance than what would be possible with an army of data scientists.
This promise has already resulted in several high-profile contracts, including seven-figure deals with Fortune 100 clients. The company has also entered a strategic partnership with AWS that will allow AWS users to deploy Nexus directly from their existing instances.