Skip to main content

Cartesia

Category: Voice / Speech AI

A real-time multimodal AI company developing ultra-low latency generative voice models powered by novel State Space Model (SSM) architectures. Cartesia was founded in 2023. The company is led by Karan Goel. Based in San Francisco, USA. Team size: 40-50. Total funding raised: $91M. Latest round: Series A. Key investors include ["Kleiner Perkins","Index Ventures","Lightspeed","General Catalyst","A*","Factory","Greycroft","Conviction","SV Angel","Dell Technologies Capital","Samsung Ventures"].

Founded
2023
Headquarters
San Francisco, USA
Team size
40-50
Total funding
$91M

Value proposition

Delivers human-like, emotionally expressive speech with industry-leading ultra-low latency (sub-100ms), enabling truly natural real-time verbal interaction between humans and AI.

Products and solutions

see above

Unique value

Unlike most competitors using Transformer-based architectures, Cartesia utilizes State Space Models (SSMs) which allow for faster-than-real-time processing and lower compute costs.

Target customer

Enterprise developers, AI agents/robotics companies, gaming studios, customer service platforms, and content creators.

Industries served

["Customer Experience & Support (AI Call Centers)","Gaming & Interactive Media","Education & EdTech","Assistive Technology & Accessibility","Content Creation & Localization"]

Technology advantage

Leverages the 'Mamba' architecture and subquadratic scaling to achieve significantly lower latency and higher efficiency than traditional models, making it the premier choice for real-time conversational AI.

How they differentiate

Utilizes State Space Models (SSMs) and the Mamba architecture instead of traditional Transformers, enabling sub-100ms latency and significantly lower compute costs for real-time applications.

Main competitors

["ElevenLabs","Deepgram","OpenAI (Voice Engine)","Hume AI"]

Key partnerships

["Together AI: Dedicated model partner - Sonic as native TTS endpoint on Together AI Voice Platform.","Rasa: Voice partner for enterprise-grade voice agents.","Cresta: Powering next-gen voice virtual agents.","NVIDIA (NVentures): Strategic investor and compute partner.","Dell Technologies Capital: Investor and infrastructure collaborator.","Samsung Ventures: Strategic investor for consumer electronics integration."]

Notable customers

["Quora","Cresta","Rasa","Thoughtly","Lorikeet","Together AI","Poe"]

Major milestones

["Founded in 2023 by researchers from Stanford AI Lab","Raised $27M Seed round led by Index Ventures (Dec 2024)","Raised $64M Series A led by Kleiner Perkins (Mar 2025)","Launched Sonic 2.0 with Series A (Mar 2025), later Sonic-3 (late 2025)","Launched Line: Modern Voice Agent Development Platform (Aug 2025)","Launched Ink: streaming speech-to-text model","Achieved GDPR compliance (Sep 2025)","Pioneered commercial application of Mamba/SSM architectures in generative voice"]

Growth metrics

50,000+ customers; Sonic used by 10,000+ customers as of Mar 2025; 40+ languages; 99.9% uptime; SOC-2 and HIPAA compliant.

Market positioning

Infrastructure-level provider of real-time multimodal AI, targeting developers of AI agents, robotics, and interactive media.

Geographic focus

Global (Headquartered in San Francisco, USA)

Patents and IP

Proprietary IP based on foundational research in State Space Models (SSMs) and the H3 architecture developed by the founders at Stanford AI Lab.

About Karan Goel

Karan Goel is a distinguished AI researcher and entrepreneur. He earned his PhD from Stanford University's AI Lab under Professor Chris Ré, where he co-authored foundational research on State Space Models (SSMs) and the H3 architecture, which paved the way for the Mamba architecture. His professional career includes roles as a Research Scientist at Snorkel AI and a Machine Learning Researcher at Salesforce AI Research. He was also a Greylock X Fellow and a Research Fellow at Microsoft Research.

Official website: