Ted Xiao
Founding Member of Technical Staff, Project Prometheus
Reviewed Updated Jul 23, 2026This profile is AI-generated. If you spot an error, please help us fix it by sharing a URL to the correct information.
Background
Ted Xiao is a Founding Member of Technical Staff at Project Prometheus, Jeff Bezos’s AI startup building AGI for the physical economy 12. He joined Prometheus in late 2025 after eight years at Google DeepMind, where he was a Staff Research Scientist and Tech Lead on the Robotics team and the Gemini Multimodal team 13. He publicly announced his departure from Google DeepMind on X in late 2025, writing that he had helped “transform the dream of general-purpose robot learning from a heretical fringe idea into a normalized technology roadmap” 3.
Xiao holds a B.S. and M.S. in Electrical Engineering and Computer Science from UC Berkeley (2013-2017), where he was advised by Professor Claire Tomlin and focused on control systems, intelligent systems, and robotics 14. As an undergraduate he founded Machine Learning at Berkeley 14. His notable research contributions include SayCan, RT-1, RT-2, Open X-Embodiment (IEEE ICRA 2024 Best Paper Award), the Octo generalist robot policy, and the Gemini Robotics series 1. He writes the Action-driven Intelligence Substack about foundation models, robotics, and decision-making 5.
Project Prometheus, co-founded by Jeff Bezos and Vik Bajaj, raised $6.2 billion at launch in late 2025 and a $12 billion Series B at a $41 billion valuation in June 2026, with approximately 150 employees across San Francisco, London, and Zurich 2.
Stated Thesis
Xiao has not published a formal investment thesis. His public writing focuses on the technical and economic significance of physical AI. On his Substack, he has argued that “physical AGI will be the final frontier of AI as a whole” and that the future “will be a future built, tested, and ultimately measured against the physical world” 6. He describes his research agenda as “scaling frontier models for real-world understanding and interaction, with a particular focus on approaches that can leverage internet-scale foundation models and methods that improve with more experience” 1.
Inferred Thesis
Only one angel investment has been publicly confirmed, so any inference from portfolio behavior is preliminary and should be treated as directional rather than statistical.
- Sector: Physical AI / embodied AI / robotics data infrastructure. The single confirmed investment (Mecka AI) trains robots on human motion data captured via body sensors and iPhones 78 — directly adjacent to Xiao’s own research area of scaling robot learning via internet-scale data.
- Stage: Series A (Mecka AI’s round closed November 2025, with a follow-on that brought the total to $60M announced June 2026) 78.
- Geography: Mecka is New York-based; Xiao himself is Bay Area-based 74.
- Co-investor pattern (n=1): Framework Ventures (lead), Menlo Ventures, SV Angel, Kindred Ventures 78.
- Founder profile: The Mecka founding team notably came from outside traditional AI — Josh Gao and Mogen Cheng previously built a restaurant payments startup, Jason Chong sold a startup to Coinbase, Duy Nguyen came from shoe reselling 7. This suggests Xiao is comfortable backing operator-founders rather than requiring an AI research pedigree.
- Check size: Not disclosed.
Sample size is one. Additional angel investments may exist but are not publicly confirmed as of research date.
Portfolio
| Company | Year | Stage | Source |
|---|---|---|---|
| Mecka AI | 2025 | Series A (participated as angel) | 78 |
Only one angel investment could be independently verified. The Mecka AI Series A closed in November 2025 with a $35M follow-on later, publicly announced June 1, 2026 78.
In Their Own Words
On leaving Google DeepMind to join Project Prometheus (late 2025):
“After 8 unforgettable years, I have decided to leave Google DeepMind. I feel immensely grateful to have had the opportunity to help transform the dream of general-purpose robot learning from a heretical fringe idea into a normalized technology roadmap.” 3
On why physical AI is the ultimate frontier (from his Substack essay “When Bits Hit Atoms,” October 2025):
“physical AGI will be the final frontier of AI as a whole” 6
“the physical world is rate limiting” and “real-world wall time cannot be hacked or gamed.” 6
What Founders Say
No independently sourced founder testimonials found. Mecka AI founders have publicly discussed their company’s vision but have not been quoted specifically about working with Xiao as an angel investor in coverage reviewed for this profile.
Sources
-
Ted Xiao personal website (tedxiao.me), “Ted Xiao — Prometheus,” accessed July 2026. https://tedxiao.me/↩↩↩↩↩↩
-
TechCrunch, “Jeff Bezos’s Prometheus raises $12B to build an ‘artificial general engineer’ for the physical world,” June 11, 2026. https://techcrunch.com/2026/06/11/jeff-bezoss-prometheus-raises-12b-to-build-an-artificial-general-engineer-for-the-physical-world/↩↩
-
Ted Xiao (@xiao_ted) on X, announcement of departure from Google DeepMind, accessed July 2026. https://x.com/xiao_ted/status/19869068915415329↩↩↩
-
Ted Xiao LinkedIn profile, accessed July 2026. https://www.linkedin.com/in/tedxiao/↩↩↩
-
Ted Xiao, “Action-driven Intelligence” Substack, accessed July 2026. https://agentic.substack.com/↩
-
Ted Xiao, “When Bits Hit Atoms,” Action-driven Intelligence Substack, October 2025. https://agentic.substack.com/p/when-bits-hit-atoms↩↩↩
-
Fortune, “Mecka AI raises $60 million to train robots with human data sourced from body sensors and iPhones,” June 1, 2026. https://fortune.com/2026/06/01/mecka-ai-series-a-60-million-robotics-data-training/↩↩↩↩↩↩↩
-
The SaaS News, “Mecka AI Raises $60M Series A,” June 2026. https://www.thesaasnews.com/news/mecka-ai-raises-60m-series-a/↩↩↩↩↩