Skip to main content

SambaNova Systems

Category: AI Infrastructure

Provides a full-stack AI platform, including proprietary hardware (Reconfigurable Dataflow Units - RDUs) and software (SambaNova Suite), to run large-scale AI and foundation models for enterprises. Offered as a subscription service (Dataflow-as-a-Service). SambaNova Systems was founded in 2017. The company is led by Rodrigo Liang. Based in Palo Alto, California, USA. Team size: 501-1000. Total funding raised: $1.13 billion. Latest round: Series D ($676M) on April 13, 2021. Key investors include ["SoftBank Vision Fund 2","BlackRock","Google Ventures (GV)","Intel Capital","Walden International","Temasek","GIC"].

Founded
2017
Headquarters
Palo Alto, California, USA
Team size
501-1000
Total funding
$1.13 billion

Value proposition

Delivers best-in-class performance and efficiency for AI training and inference through its purpose-built, reconfigurable dataflow architecture. This offers a faster, more secure, and more cost-effective solution for enterprises to own and deploy their own foundation models compared to GPU-based alternatives.

Products and solutions

["SambaNova Suite (featuring Composition of Experts and Deep Research Agents)","DataScale Systems (Hardware with RDUs)","Dataflow-as-a-Service (DaaS)","Samba-1 (1T parameter model)"]

Unique value

Its custom-built Reconfigurable Dataflow Unit (RDU) is designed from the ground up for AI data flow, offering superior efficiency and performance for training and inference of large, complex models compared to repurposed GPUs.

Target customer

Large enterprises and government organizations requiring high-performance, scalable, and secure AI infrastructure for developing and deploying foundation models and generative AI applications.

Industries served

["Financial Services","Healthcare & Life Sciences","Government & Public Sector","Manufacturing","Research & High-Performance Computing"]

Technology advantage

A full-stack, integrated hardware and software solution (Dataflow-as-a-Service) optimized for large foundation models, providing a faster, more secure, and more cost-effective alternative to traditional GPU-based systems for enterprises.

How they differentiate

Differentiates through its unique Reconfigurable Dataflow Unit (RDU) architecture, which is software-defined and optimized for dataflow in AI models, unlike the fixed architecture of GPUs. This full-stack approach (hardware and software) allows for higher efficiency and performance on complex, large foundation models.

Main competitors

["Nvidia","Cerebras Systems","Groq","Graphcore"]

Key partnerships

["AWS","SoftBank","Oracle","Hugging Face","Lepton AI"]

Notable customers

["US Department of Energy's Argonne National Laboratory","OTP Group","SoftBank Corp","Fortune 500 companies in finance, healthcare, and government sectors"]

Major milestones

["Secured over $1.13 billion in total funding","Achieved a valuation of over $5.1 billion after its Series D round","Launched Samba-1, one of the first trillion-parameter models for enterprise use","Appointed semiconductor industry veteran Lip-Bu Tan as Executive Chairman"]

Growth metrics

Recognized as one of Fast Company's Most Innovative Companies of 2024.

Market positioning

A key challenger to Nvidia's dominance, positioned as a high-performance, specialized full-stack solution for enterprises deploying large language and foundation models at scale.

Geographic focus

Global, with a primary focus on North America, Europe, and Asia.

Patents and IP

Holds numerous patents for its core Reconfigurable Dataflow Architecture, dataflow processing systems, and methods for implementing operations in AI hardware.

About Rodrigo Liang

Rodrigo Liang is the co-founder and CEO of SambaNova Systems, a role he has held since the company's inception in November 2017. He has over two decades of experience in the semiconductor and enterprise systems industry.

Official website: