Kaon AI, a generative AI platform and research lab building interactive entertainment experiences, has closed a Series B round backed by B Capital, Redpoint Ace, Goodwater Capital and DCM. The funding, whose size is undisclosed but reported by Variety and others to be around $60 million, will fuel Kaon’s push to define a new content engine for the generative AI era, built on long‑horizon, hyper‑personalized narratives rather than static recommendations.
While Big Tech focuses on automating the workplace, Kaon is targeting what it sees as the most valuable commodity of the generative era: consumer attention. Led by CEO Jay Dang, CTO Alex Xi and COO Lifan Wang, the company’s flagship consumer app Emochi already has more than 2 million daily active users who spend an average of 150 minutes per day on the platform. Emochi alone is said to generate about $45 million in annual recurring revenue, contributing to an eight‑figure ARR across Kaon’s AI‑native ecosystem, which also includes FlowGPT, Tickle and Branches. By positioning deep computer science as a vehicle for mainstream digital culture, Kaon is achieving user engagement metrics far beyond most general‑purpose AI tools.
Rather than simply lowering the cost of content production, Kaon’s Berkeley‑educated founding team is focused on changing the nature of content itself. Their thesis: as AI drives production costs toward zero, the real challenge becomes making content that responds to the individual consuming it, turning passive viewers into co‑creators. Kaon’s proprietary architecture operates as a “customization engine” that writes, illustrates and evolves immersive multimedia narratives frame by frame based on real‑time user behavior, blending story generation, character development and interactive world‑building into a continuous loop.
To make this model economically viable at scale, Kaon has built a full‑stack inference infrastructure that manages more than a thousand GPUs across partners such as Nebius and DigitalOcean. The system is designed to handle trillion‑level daily token volumes at roughly 10 times lower cost than traditional cloud architectures, giving Kaon room to offer rich, persistent experiences without unsustainable compute spend. Investors say the team’s insistence on owning the token‑to‑application stack—from front‑end interactions and in‑app economies down to core inference infrastructure—creates a compounding advantage in performance, data and iteration that is difficult for more “partner‑oriented” platforms to replicate.
The Series B will also help scale Kaon Labs, the company’s research arm, which will use Kaon’s millions‑strong, highly engaged consumer base as a live sandbox for rapid model experimentation. With end‑to‑end control over the ecosystem, the founders aim to deploy and test new architectures in days rather than months, tightening the feedback loop between research and production and pushing the frontier of interactive AI entertainment. Kaon is expanding its R&D footprint in San Mateo and deepening ties with academic advisors and collaborators, drawing on its founders’ UC Berkeley roots.
KEY QUOTES:
“We believe interactive AI entertainment is emerging as the next distinct consumer category. Kaon stands out because Jay, Alex, and Lifan have built a full-stack powerhouse that owns its data pipeline from the ground up. That level of integration creates a compounding advantage in performance, data, and product iteration that is difficult to replicate, and we are excited to support the Kaon team as they help define this new category.”
Daisy Cai, General Partner, B Capital
“Every streaming platform today recommends what could be your next favorite show from a long list of preset options. In the generative AI era, though, this finite catalog is poised to become the ‘Blockbuster’ of the 2020s. At Kaon, we’re building a new type of content engine to create individualized narratives that truly resonate, moving the industry past static text prompts and into real-time multimedia.”
Jay Dang, CEO and Co‑Founder, Kaon AI
