An AI startup called Fundamental just emerged from stealth with a jaw-dropping $255 million war chest and a bold claim: it's built the first foundation model that can actually make sense of the massive structured datasets sitting inside enterprises. The company's Nexus model takes a completely different approach than OpenAI or Anthropic, ditching the transformer architecture entirely to tackle spreadsheets with billions of rows - something contemporary LLMs struggle with. With Fortune 100 contracts already signed and a strategic AWS partnership in place, Fundamental thinks it's found the missing piece in the enterprise AI puzzle.
Fundamental just threw down the gauntlet in enterprise AI with a $255 million Series A that dwarfs most startups' entire funding trajectories. The company emerged from stealth Thursday with a provocative pitch: while everyone's been chasing the next ChatGPT, they've been quietly building something enterprises actually need - a foundation model that can chew through the billions of rows of structured data sitting in corporate databases.
"While LLMs have been great at working with unstructured data, like text, audio, video, and code, they don't work well with structured data like tables," CEO Jeremy Fraenkel told TechCrunch. "With our model Nexus, we have built the best foundation model to handle that type of data."
The funding round tells you everything about investor appetite for this approach. Oak HC/FT, Valor Equity Partners, Battery Ventures, and Salesforce Ventures co-led the massive Series A, with Hetz Ventures joining and notable angels including Perplexity CEO Aravind Srinivas, Brex co-founder Henrique Dubugras, and Datadog CEO Olivier Pomel throwing their weight behind it. That's $225 million in the Series A alone, with an additional $30 million in earlier seed funding bringing the total to $255 million.
What makes Nexus different isn't just marketing spin. The model is what Fundamental calls a Large Tabular Model (LTM) rather than a Large Language Model, and it breaks from the transformer architecture that defines everything from GPT-4 to Gemini. Instead of the probabilistic outputs that make LLMs occasionally hallucinate, Nexus is deterministic - ask it the same question twice, and you'll get the same answer both times.
