In December 2023, Liquid AI, an MIT spin-off co-founded by CSAIL scientists, raised $46.6 million for the development of liquid neural networks, smaller but potentially more capable AI models that require less computing power to run. This significant seed round marked the beginning of the Liquid AI project.
In December 2024, Liquid, an AI startup, raised $250 million in an early-stage funding round led by AMD. The company aims to build the most capable and efficient AI system at every scale by 2025, with more investments expected.
Liquid AI: What is it?
Unlike conventional transformer-based AI, or GPTs, Liquid AI, an MIT spin-off, creates generative AI models using a fundamentally different architecture.
Ramin Hasani, Mathias Lechner, Alexander Amini, and Daniela Rus are the four co-founders of the startup. They all have exceptional backgrounds in machine learning (ML) and AI research.
Liquid AI Co-Founders
- Daniela Rus, MIT’s co-founder and director of Liquid AI, is a professor of electrical engineering and computer science, with research interests in robotics, mobile computing, and data science.
- Ramin Hasani, co-founder and CEO of Liquid AI and CSAIL machine learning research affiliate, specializes in robust deep learning and decision-making in complex dynamical systems.
- Mathias Lechner, co-founder and CTO of Liquid AI and Research Affiliate at CSAIL, focuses on creating robust and reliable machine-learning models.
- Alexander Amini, co-founder and CEO of Liquid AI, leads research on autonomous science and engineering, focusing on safe decision-making for autonomous agents.
Liquid AI’s Pioneering Research: The Start of New AI?
Liquid Neural Network: What Is It?
Liquid neural networks are brain-inspired machine learning systems designed to remain adaptable and robust even after initial training, as per Liquid AI.
Liquid neural networks, expressive continuous-time machine learning systems, are capable of modeling long-term dependencies in sequential data, but their differences from traditional AI models are not fully understood.
Liquid neural networks have a dynamic architecture that adapts to new data, allowing real-time learning. They are better suited for handling changing situations and time-series data, as they are smaller and simpler. They are also easier to understand and work with fewer parameters, requiring fewer computational resources.
What is a liquid foundation model (LNN to LFM)?
The team introduced liquid foundation models (LFMs), a new generation of generative AI models, with the LFM-7B model, released on January 20, 2025, being the best-in-class language model in English, Arabic, and Japanese for private enterprise chat, code, fast instruction following, and agentic workflows.
Liquid AI highlights the LFM-7B’s key capabilities, including expansive knowledge, optimization for private enterprise chat, coding, fast instruction following, and agentic workflows, a minimal memory footprint, and superior size class performance.
The STAR Architecture
A Scalable Transformer Alternative Representations (STAR) architecture, which attempts to increase AI efficiency and scalability, is another significant component of Liquid AI’s innovation that merits special note.
Liquid AI claims that a new STAR model architecture performs better than Transformers. In 2017, Google researchers unveiled Transformer, the technology that powers the majority of today’s generative AI models.
Is It Possible to Invest in Liquid AI Stock?
No, Liquid AI is a privately held corporation, and its stock is not traded publicly.
The company has not yet made any formal declarations regarding its intentions to go public, despite certain whispers that have been going around about a possible IPO.
Liquid AI Funding: $297 Million
Over the course of two rounds, Liquid AI has raised $297 million in funding.
The firm will be able to expedite the deployment of its models across a variety of real-life scenarios in e-commerce, biotechnology, financial services, and telecommunications thanks to the most recent $250 million Series A funding round, which was headed by AMD, a market leader in AI GPUs.
The Bottom Line
Daniela Rus, the director of Liquid AI, concludes by saying, “Today’s AI has a ceiling.Let’s not accept what is now available.
