Key Takeaways
- SambaNova launched SambaManaged, a rapid deployment AI inference solution that can be operational in just 90 days.
- The modular design minimizes infrastructure changes and effectively handles increasing AI demands while reducing costs.
- Companies can choose between fully managed or self-service options, allowing flexible operational control as expertise grows.
Modular and Efficient AI Infrastructure
PARIS–SambaNova has introduced SambaManaged, the first inference-optimized data center product designed for rapid deployment. With a launch timeline of just 90 days—significantly shorter than the typical 18 to 24 months—SambaManaged allows data centers to implement AI inference services without extensive infrastructure overhauls.
As global demand for AI inference continues to rise, traditional data centers face challenges such as long deployment timelines, high power consumption, and expensive facility upgrades. SambaManaged aims to overcome these obstacles, enabling organizations to quickly roll out profitable AI inference services by utilizing existing power and network structures.
According to Abhi Ingle, Chief Product and Strategy Officer at SambaNova, data centers are struggling with issues related to power, cooling, and expertise as AI demand grows. “SambaManaged delivers high-performance AI with just 10kW of air-cooled power and minimal infrastructure changes, making rapid deployment simple for any data center,” he noted.
Key Advantages of SambaManaged:
-
Unmatched Efficiency: SambaManaged sets a new benchmark for performance per watt, helping organizations maximize their return on investment while lowering overall costs.
-
Rapid Deployment: The service can be operational in as little as 90 days, minimizing integration challenges and accelerating the path to value.
-
Open Model Flexibility: Organizations can utilize leading open-source models for rapid inference, avoiding vendor lock-in and ensuring their operations are future-proof.
-
Modular, Scalable Design: The product allows for easy scaling from small to large deployments, including options to build a “Token Factory” with up to 1 MW (100 racks or 1,600 chips), accommodating evolving business needs.
-
Managed and Self-Service Options: Companies have the flexibility to choose a fully managed service or transition to self-service as internal capabilities develop, supported by a customizable UI and various pricing models.
SambaManaged is already in use by a major US public company with significant power demands, optimizing revenue through top-throughput capabilities on models like DeepSeek and improving Power Usage Effectiveness (PUE).
Rodrigo Liang, CEO and co-founder of SambaNova, emphasized the immediacy of their solution, stating, “While others talk about the future of AI, we’re delivering it today. SambaManaged is a game-changer for organizations eager to accelerate their AI initiatives without sacrificing speed, scale, or efficiency. Wherever there is power and networking, we can establish AI infrastructure swiftly.”
Founded in 2017 and headquartered in Palo Alto, California, SambaNova enables enterprises to swiftly adopt advanced generative AI technologies. The company is supported by esteemed investors, including the SoftBank Vision Fund 2 and Intel Capital.
For more details, please visit sambanova.ai or reach out to info@sambanova.ai.
The content above is a summary. For more details, see the source article.
