Today isn't just a product launch; it's the dawn of a new chapter for SambaNova. We're unveiling a bold new brand identity, a sharpened market focus, and a powerful brand positioning centered on one core belief: the relentless pursuit of intelligence.
Developers worldwide are building with relentless intelligence – creating AI applications that demand real-time inference at unprecedented scale. The demand for real-time AI inference is exploding, reshaping industries and redefining competitiveness. However, this surge is colliding head-on with the physical and operational realities of modern data centers. Legacy infrastructure and traditional compute approaches buckle under the overwhelming power, cooling, and expertise demands of high-performance AI. Data centers face a critical inflection point: adapt swiftly to deliver these essential AI services or risk losing relevance as customers seek providers who can.
Recognizing this pivotal moment, SambaNova has transformed. We've emerged as a modern, agile innovator, laser-focused on delivering the most efficient and adaptable AI platform on the planet.
Our North Star: Sovereignty Through Efficiency
- Vision: To be the most efficient and adaptable AI platform on the planet – so that every enterprise can confidently, securely and privately, own the trajectory of their data and AI future.
- Mission: To provide every developer, enterprise, government, and data center absolute sovereignty over their data, models, and AI infrastructure – future-proofing the workloads powering tomorrow.
- Positioning: The Relentless Pursuit of Intelligence. We deliver high-efficiency, low-power, chips-to-models computing, enabling humans and organizations to achieve extraordinary intelligence.
Product Portfolio to Meet Our Customers Where They Are
Our new offering is three products that share one common system and platform: SambaCloud, SambaStack, and SambaManaged - all powered by the same award-winning AI chip, the SN40L.
The Architecture of Relentless Intelligence
System: Where relentless intelligence takes physical form
At the heart of SambaNova innovation is the Reconfigurable Dataflow Unit (RDU). Purpose-built for AI workloads, the RDU takes advantage of a dataflow architecture and a three-tiered memory design. The three tiers of memory enable the platform to run hundreds of models on a single node and to switch between them in microseconds. The Dataflow Architecture allows data between AI operations to flow seamlessly through the chip due to the unique layout of compute and memory. These advantages combined is what ultimately results in the efficiency and performance gains experienced when running AI workloads.
SambaRack contains 16 SN40L RDUs with the necessary software and hardware components necessary to run inference on the largest AI models. Each rack is optimized for power consumption and can be air-cooled to work within existing data centers in a small footprint.
Platform: Software that makes intelligence accessible and performant
SambaOrchestrator is our common Kubernetes orchestration platform across all our products that dynamically manages AI workloads across SambaRack. With one common platform, customers can easily leverage the performance and efficiency of our system and scale to meet the ever-growing demand for AI.
These AI inference workloads are made accessible by a common set of APIs that are based on industry standards. This allows developers to easily port their existing applications to SambaNova and get started building in minutes.
Products: Accelerating every customer’s AI journey
Our product offerings are specifically designed to meet customers where they are at in their AI journey. Whether you are a developer just getting started building or a data center company that is looking to get into the AI market, SambaNova has offerings that help you where you are and allow you to scale.
SambaCloud delivers fast inference on the best open-source models, powered by SambaRack in our data centers. Developers and enterprises can get started building in minutes with models like DeepSeek or Llama 4. Fast inference is especially important for the wave of agents we are starting to see emerge that require many different models in order to turn insights into actions. See the demo of our Demo Research Agent that leverages models like DeepSeek to search the web and can now produce comprehensive reports in various formats.
SambaStack: Enterprises, governments, and data centers require dedicated infrastructure for privacy, security, and sovereignty. SambaStack allows these organizations to purchase dedicated hardware, either hosted in the cloud or on-premise, to run their most sophisticated AI workloads. Because of the common platform, it is easier than ever to start on SambaCloud and scale into SambaStack.
NEW SambaManaged: Launching today, we are now offering a fully managed turnkey inference service for data centers, SambaManaged. This product makes it easier than ever for enterprises and data centers to create their own AI inference cloud using their existing data center infrastructure and power. Because SambaRacks are air-cooled and produce the most tokens per watt of energy, data center customers can utilize existing infrastructure and stand up a service in just a matter of 90 days.
Read more about the offering in our launch blog.
The Future is Efficient. The Future is Now.
While others speculate about the future of AI, SambaNova is delivering it today. Our rebrand reflects our unwavering commitment to the relentless pursuit of intelligence through extreme efficiency and adaptability. SambaManaged embodies this commitment, providing the fastest path to sovereign, high-performance AI inference for the modern data center.
Ready to own your AI future with relentless efficiency?
SambaNova: The Relentless Pursuit of Intelligence.