Rafay
Category: AI Infrastructure
Rafay Systems builds an infrastructure orchestration platform that lets neoclouds, sovereign AI clouds, telcos and enterprises turn GPU and CPU infrastructure into self-service, governed AI and cloud services. Rafay was founded in 2017. The company is led by Haseeb Budhani. Based in Sunnyvale, California, USA. Team size: 101-500. Total funding raised: $37.1M. Latest round: Series B. Key investors include ForgePoint Capital, Ridge Ventures, Costanoa Ventures, Moment Ventures, NTT DOCOMO Ventures.
- Founded
- 2017
- Headquarters
- Sunnyvale, California, USA
- Team size
- 101-500
- Total funding
- $37.1M
Value proposition
Turns raw GPU/CPU capacity into governed, multi-tenant, self-service, monetizable AI cloud services (Kubernetes, SLURM, VMs, bare metal, Token Factory inference), replacing ticket queues and manual provisioning.
Products and solutions
Rafay Platform (GPU and CPU infrastructure orchestration), AI Factory, Token Factory (token-metered model inference), Token Delivery Networks, Kubernetes management and operations, as-a-Service offerings (SLURM, VMs, Kubernetes clusters, bare-metal GPUs), GPU PaaS Reference Architecture, open-source projects and ecosystem integrations.
Unique value
One orchestration layer that lets operators launch CSP-grade, multi-tenant AI cloud services (including Token Factory) on their own GPU infrastructure, with governance, metering, quotas and audit trails built in.
Target customer
Neoclouds, sovereign AI cloud providers, telcos and cloud service providers, and large enterprises running AI/ML or cloud-native workloads.
Industries served
Neoclouds/GPU clouds, sovereign AI clouds, telecommunications, enterprises (financial services, healthcare, technology)
Technology advantage
NVIDIA AI Cloud Ready validation; co-authored GPU PaaS Reference Architecture with NVIDIA; Cisco Solutions Plus partnership (procurable on Cisco purchase orders); confidential AI orchestration for CNCF Confidential Containers via Fortanix; the platform covers Kubernetes, SLURM, VMs and bare metal.
How they differentiate
Rafay focuses on the orchestration and service-delivery layer for neoclouds and sovereign AI (self-service, multi-tenancy, metering, Token Factory) rather than only Kubernetes cluster management, and it works closely with NVIDIA and Cisco.
Main competitors
Spectro Cloud, Red Hat OpenShift, Fluidstack
Key partnerships
NVIDIA (AI Cloud Ready validation, GPU PaaS Reference Architecture), Cisco (Solutions Plus partner since June 2026, Nexus One and Secure Firewall integrations), Fortanix (confidential AI on Kubernetes pods via CNCF Confidential Containers), Accenture and NVIDIA (sovereign AI whitepaper), Netris (GPU cloud monetization)
Notable customers
TELUS (sovereign AI Studio), SonicWall
Major milestones
Founded 2017, $4.1M seed, $8M Series A and SaaS GA in Sept 2019, $25M Series B in Aug 2021, NVIDIA AI Cloud Ready validation, Cisco Solutions Plus partner in June 2026, Fortanix confidential AI on Kubernetes pods announced in Oct-Nov 2026.
Growth metrics
Estimated 2024 revenue of about $25.1M, up from $6.8M in 2023 (Latka estimate, unverified); about 158 employees per Latka. A $4.1M seed round preceded the Series A. Trackers such as Tracxn and Crunchbase list about $33M across the two disclosed priced rounds.
Market positioning
AI infrastructure orchestration software for neoclouds, sovereign AI clouds, telcos and enterprises, which began as a Kubernetes operations platform and has moved up the stack into AI cloud and token monetization.
Geographic focus
Global (US, Canada, and sovereign AI markets worldwide)
Patents and IP
U.S. Patent US10791168B1, “Traffic aware network workload management system,” granted September 29, 2020; original assignee: Rafay Systems, Inc.
About Haseeb Budhani
Co-founder and CEO of Rafay Systems since 2017; previously spent a year at Akamai; advisor to Saaras Inc. and Cyberhaven.
Latest news about Rafay
More AI Infrastructure companies
Official website: https://rafay.co