Elorian
Category: Foundation Models / LLMs
A stealth-mode AI research company developing next-generation multimodal foundation models with a specific focus on advanced visual reasoning and the path toward Artificial General Intelligence (AGI). Elorian was founded in 2025. The company is led by Andrew Dai. Based in Palo Alto, USA. Team size: 11-50. Total funding raised: $55.0M. Latest round: Seed, $50.0M, 2025-01, led by Undisclosed. Key investors include Striker Ventures, Menlo Ventures, Altimeter Capital, NVIDIA, 49 Palms, Jeff Dean.
AMW Analysis
Elorian is a stealth-mode AI research company developing multimodal foundation models focused on advanced visual reasoning, aiming to bridge the gap between text-centric LLMs and true visual understanding. Founded in 2025 and based in Palo Alto, the company has raised a total of $55.0M, with its latest seed round of $50.0M in January 2025 led by undisclosed investors, including Striker Ventures, Menlo Ventures, Altimeter Capital, NVIDIA, 49 Palms, and Jeff Dean.
Recent news flow shows Elorian secured a $55M seed round at a $300M valuation, backed by NVIDIA, Menlo Ventures, and Jeff Dean, targeting visual reasoning AI for robotics, aerospace, medicine, and manufacturing. The company plans to expand its research team, boost compute resources, and launch pilots in factories. Earlier in January 2026, founders Andrew Dai and Yinfei Yang, former leaders from Google Gemini and Apple, launched the startup seeking a $50 million seed round to build native multi-modal models that process text, images, and video simultaneously, aiming for AI agents that can navigate software GUIs and physical logic.
AMW analysis, generated from 2 tracked news signals.
- Founded
- 2025
- Headquarters
- Palo Alto, USA
- Team size
- 11-50
- Total funding
- $55.0M
Value proposition
Bridging the gap between text-centric LLMs and true visual understanding, enabling AI to perform complex logical reasoning tasks based on visual inputs rather than simple image recognition.
Products and solutions
Multimodal Foundation Models (LVM/LMM), Visual Reasoning Engine, AGI Research Frameworks, Proprietary Multimodal Training Data Pipelines
Unique value
Founded by the core architects of Google's Gemini (data lead) and Apple's vision models, the company focuses on 'reasoning' within the visual domain—a known bottleneck for current frontier models.
Target customer
Enterprise AI developers, robotics companies, autonomous systems manufacturers, and high-end tech firms requiring advanced visual-spatial intelligence.
Industries served
Artificial Intelligence, Computer Vision, Robotics & Automation, Autonomous Vehicles, Enterprise Software
Technology advantage
Leverages deep expertise in scaling massive multimodal datasets and training high-efficiency vision-language models, utilizing proprietary techniques developed by the founders during their tenures at Google DeepMind and Apple.
How they differentiate
Focuses specifically on 'visual reasoning'—the logical processing of visual-spatial data—leveraging the founders' expertise in scaling Gemini's multimodal data and Apple's vision systems to solve the reasoning bottleneck in AGI.
Main competitors
World Labs, Physical Intelligence (Pi), General Intuition
Key partnerships
Striker Ventures (Lead Investor), Menlo Ventures (Co-Lead Investor), Altimeter Capital (Co-Lead Investor), NVIDIA (Strategic Investor), Jeff Dean (Advisor/Investor)
Major milestones
Founded in 2025 by core architects of Google Gemini and Apple Vision models, Entered stealth mode to develop visual reasoning engines for AGI, Raised $55M seed round at $300M valuation co-led by Striker Ventures, Menlo Ventures, and Altimeter Capital (Apr 2026), Emerged from stealth on April 9, 2026
Growth metrics
Raised $55M seed round at a $300M post-money valuation (April 2026). The round was raised in two tranches: first at $120M valuation, second at $300M.
Market positioning
Frontier AI research lab in stealth mode targeting the next generation of multimodal foundation models.
Geographic focus
North America (San Francisco-based)
Patents and IP
No registered patents disclosed as of latest update (Stealth mode).
About Andrew Dai
Andrew Dai is a veteran AI researcher who spent 14 years at Google and Google DeepMind. He was a Director and Principal Research Scientist, notably serving as the co-lead for the pre-training data of Google's Gemini model. His work has been foundational in large language models and multimodal AI reasoning.
Latest news about Elorian
- Elorian secured a $55M seed round at a $300M valuation, backed by NVIDIA, Menlo Ventures and Jeff Dean. The startup is targeting visual reasoning AI for robotics, aerospace, medicine and manufacturing
- Andrew Dai and Yinfei Yang, former leaders from Google Gemini and Apple, have launched Elorian to tackle the core issue of visual reasoning for AGI. The startup is seeking a $50 million seed round to
More Foundation Models / LLMs companies
Official website: https://elorian.ai
