# OctoAI

> OctoAI (formerly OctoML) was a Seattle-based AI infrastructure company founded in 2019 by University of Washington researchers — including Apache TVM creator Tianqi Chen and CEO Luis Ceze. The company built a generative AI inference platform that gave developers fast, affordable API access to leading open-source LLMs and image generation models, along with OctoStack, an enterprise-grade private AI deployment stack. After raising ~$132M and pivoting from ML optimization to GenAI infrastructure, OctoAI was acquired by NVIDIA in September 2024 and wound down its commercial services by October 31, 2024.

- **Founded:** 2019
- **Headquarters:** Seattle, Washington, USA
- **Founders:** Luis Ceze (CEO & Co-founder), Tianqi Chen (CTO & Co-founder), Jared Roesch (Chief Architect & Co-founder), Jason Knight (CPO & Co-founder), Thierry Moreau (VP of Technology Partnerships & Co-founder)
- **Team size:** Not publicly disclosed; startup-scale (~100–200 employees estimated at peak)
- **Products:** OctoAI SaaS Platform, OctoStack, OctoAI Media Gen Solution, OctoML Platform (legacy)
- **Notable:** Co-created Apache TVM, now one of the most widely-used open-source ML compiler frameworks in the world., Spun out of UW Allen School in 2019, commercializing academic research into a venture-backed startup., Raised ~$132M total across four rounds, reaching a ~$900M valuation at peak.

## Products & services

- **OctoAI SaaS Platform** — A cloud-based generative AI inference platform offering fast, affordable API access to leading open-source LLMs (Meta Llama family, Mistral 8x7B, etc.) and image generation models. Within 10 months of launch it served 25,000+ developers and 100+ high-growth businesses.
- **OctoStack** — Launched April 2024, described as the industry's first complete tech stack to serve generative AI models anywhere — deployable in virtual private clouds, on-premises hardware, or public clouds (AWS, Azure, GCP). Claimed 4x better GPU utilization and ~50% reduction in operational costs vs. DIY deployments. Supported NVIDIA, AMD, and AWS Inferentia accelerators.
- **OctoAI Media Gen Solution** — An image and video generation product built on Stable Diffusion models (SDXL, SD 1.5, Stable Video Diffusion), FLUX.1, and others. Offered text-to-image, image-to-image, image-to-video, inpainting, outpainting, upscaling, background removal, and ControlNet features via API, web UI, and Python/TypeScript SDKs.
- **OctoML Platform (legacy)** — The original ML model optimization and deployment product (2019–2023) built on Apache TVM. Focused on accelerating traditional ML workloads for computer vision and NLP by compiling models to run efficiently on diverse hardware.

## Achievements

- Co-created Apache TVM, now one of the most widely-used open-source ML compiler frameworks in the world.
- Spun out of UW Allen School in 2019, commercializing academic research into a venture-backed startup.
- Raised ~$132M total across four rounds, reaching a ~$900M valuation at peak.
- Onboarded 25,000+ developers and 100+ enterprises on the SaaS platform within ~10 months of the generative AI product launch.
- Claimed to offer the fastest Stable Diffusion XL inference endpoint commercially available (2024).
- OctoStack achieved 4x better GPU utilization vs. DIY deployments according to company benchmarks.
- Acquired by NVIDIA in September 2024 — NVIDIA's 5th acquisition of 2024.
- CEO Luis Ceze joined NVIDIA as VP of AI Systems Software post-acquisition.

## Latest updates

- **2024-10** — OctoAI winds down all commercial services for customers on October 31, 2024, following NVIDIA acquisition.
- **2024-09** — NVIDIA officially acquires OctoAI. CEO Luis Ceze joins NVIDIA as VP of AI Systems Software. octoml.ai domain redirects to nvidia.com.
- **2024-07** — FLUX.1 [Schnell] text-to-image model added to OctoAI Media Gen platform.
- **2024-04** — OctoStack launched — billed as the industry's first complete generative AI production stack for enterprise private deployments.
- **2024-01** — OctoML formally rebrands to OctoAI, reflecting the company's full pivot to generative AI infrastructure.
- **2023-11** — OctoAI Image Gen solution unveiled, enabling Stable Diffusion customization and inference at scale.
- **2023-06** — OctoAI generative AI SaaS product publicly launched.

## Links

- Website: https://octo.ai
- LinkedIn: https://www.linkedin.com/company/octoaicloud
- Twitter/X: https://x.com/OctoAICloud
- GitHub: https://github.com/octoml

---

Profile page: https://yespress.io/octoai
Published by YesPress — https://yespress.io
Last updated: 2026-04-03