🧔♂️ A friendly human may check it before it goes live. More news here
Ex-Google DeepMind researchers’ startup Elorian raises $55m
Elorian has come out of stealth with US$55 million in funding at a US$300 million valuation.
The Palo Alto-based startup, co-founded by former Google DeepMind researcher Andrew Dai, is building models that reason about images and other visual data.
Menlo Ventures, Altimeter Capital, Striker Venture Partners, Nvidia, and prominent AI researcher Jeff Dean backed the round, which was raised in two tranches at valuations of US$120 million and US$300 million, the company said.
Elorian is not generating revenue yet, but Dai said it is in talks with potential customers and plans to release its first public reasoning model in around 12 months.
Other co-founders include former Google and Apple researcher Yinfei Yang and former Harvard professor Seth Neel.
🔗 Source: Bloomberg
🧠 Food for thought
Implications, context, and why it matters.
Elorian ties its mission to visual reasoning, the physical world, and longer-term artificial general intelligence (AGI) ambitions
- Elorian wants to go past basic image analysis and pursue what it calls the “core path to AGI” 1.
- Its plan centers on models that grasp the physical world, including spatial relationships and physical constraints 2.
- That focus places the company in the “physical AI” market, which builds intelligence for robots and autonomous systems 3.
- Co-founder Andrew Dai previously worked at Google DeepMind, Google’s AI research unit. Some reporting also says he helped lead data-focused pre-training for Google’s Gemini models 4.
Momentum continues toward AI that can perceive and operate in the physical world
- Elorian’s launch adds to a shift of talent and capital toward AI that can perceive and operate in the physical world, with uses such as robotics and autonomous systems 3.
- The company is building an intelligence layer for robotics by training AI models instead of making robots, a playbook seen in other high-valuation startups 5.
- Senior researchers have left places like Google DeepMind and Apple to start similar labs. A belief in faster progress with smaller teams may be part of the story, though the provided source material does not establish this motivation 1.
- This fits the push toward multimodal models that handle text, images, video, and audio for real-world problems 4.
Stay updated on the go with our mobile app.
Get latest insights with smoother, more personalized experience through TIA mobile app.




