Holistic AI, a French startup working to surpass other generative AI models, has closed the first tranche of $200 million in an initial financing round, according to people familiar with the deal. According to Bloomberg, investors agreed to put in $80 million in equity and $120 million in convertible debt to buy computing power. It notes that the funding round values the company at $370 million.
Venture firm Accel Partners LP is participating in the round along with UiPath Inc. Other investors include Wendy and Eric Schmidt, Google’s former chief executive officer.
Accel, a fund based in California, was an early backer of Facebook, Dropbox, and several other prominent internet companies. But the firm has backed startups building foundational AI models, such as OpenAI and Mistral, a Parisian company to reach a valuation of over $2 billion in its first year. Accel also backed UiPath, a company formed in Romania that went public in 2021.
Founded by DeepMind alumni
Formed by a group of scientists who recently left DeepMind, Google’s AI lab, Holistic AI has ambitious plans. The company was founded by Charles Kantor, a former venture resident at Stanford University’s business school. The other four founding team members are engineers who spent years at DeepMind working on some of its advanced models.
It’s working to develop “multi-agent” models, a process where a number of AI systems interact with one another. The deal for the months-old startup marks continued intense interest in companies pushing forward in AI, with investors willing to take a bet on founding teams based on their employment history as much as their proposed product.
In an investor presentation, the startup described its goal to develop AI with skills like memory, planning, and “tactical decision making” that AI systems have struggled with, according to a document viewed by Bloomberg News. “We build genuine AGI capabilities,” read the presentation, referencing theoretical AI systems that outperform humans on all tasks.
In this document, Holistic AI pitched a vision of moving beyond large language models like OpenAI’s GPT and Google’s Gemini. The document said the startup was planning to target “multiple verticals” and revenue streams. “We are raising a large round to train the model and start going after these markets,” the presentation read.
