StartupWiki is a community-driven AI-powered research directory of global startup ventures. Explore in-depth profiles of AI, Biotech, CleanTech, FinTech, Cybersecurity, and other deep-tech startups with funding data, financial metrics, competitive analysis, and team insights.
Browse categories: AI & Machine Learning, FinTech, Biotech, Cybersecurity, CleanTech, Quantum Computing, and more.
Browse all startups A-Z for the complete directory.
Read our blog for startup insights and deep dives.
Real-time voice AI built on state space models — the S4/Mamba researchers' bet on instantaneous speech
HQ: San Francisco, CA, United States | Founded: 2024 | Employees: ~50 (est.) | Stage: Series B ($6.6B tender valuation, Nov 2025) | Website: https://www.cartesia.ai
Cartesia was founded in 2024 by a team of Stanford AI Lab researchers who literally wrote the literature their company runs on: Karan Goel and Brandon Yang co-authored S4ND and the S4 line of work, Albert Gu is the lead author of S4 and the main author of Mamba, and Arjun Desai rounds out the founding research team. Their thesis: transformers dominate language, but real-time voice is a different physics problem — it demands constant low latency, unbounded streaming context, and small memory footprints, exactly where recurrent-style state space models outperform attention.
The product line is built for the voice-agent era. Sonic, the flagship text-to-speech model family, generates natural speech in near-real time with full control over voice, speed, and emotion; Sonic 2.0 shipped alongside
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki