Earlier this week, SambaNovaannounced that SoftBank had become its first investor in the company’s $676 million Series D round. The funding will help SambaNova scale up its product line, including DataScale, a cloud-based platform for managing massive amounts of data. As a result, the company expects that its new products will help improve the speed and accuracy of analytics and decision-making.
SoftBank’s $676 million Series D investment
Earlier this month, SoftBank’s $676 million Series D investment in Sambanova ai, a startup that develops hardware and software for artificial intelligence applications, closed. The round was led by SoftBank Vision Fund 2, which includes GV (formerly Google Ventures) and Temasek Holdings, as well as BlackRock, Intel Capital and Singapore sovereign wealth fund GIC.
SambaNova’s main offering is Dataflow-as-a-Service, which enables enterprises to tap into the SambaNova AI system without having to invest in the hardware or maintain it. In addition to deep learning programs, the company also offers support for building large natural language processing models. It is designed to help organizations jumpstart their AI initiatives.
SambaNova plans to use its new funding injection to challenge legacy competitors. It will focus on developing and selling software-based systems for enterprises. The company has developed a scalable systems architecture, which avoids memory bottlenecks and other factors that impact traditional CPU and GPU architectures.
The company’s flagship chip, called SambaNova DataScale, is a software-defined hardware system that includes a SambaFlow development interface, which is optimized to work with open source machine learning frameworks. It is being tested with the Department of Energy and is available for sale or lease.
Accolades throughout the AI industry
Founded by Silicon Valley veterans and world-class hardware and software experts, SambaNova Systems delivers a comprehensive platform for AI. This includes Dataflow-as-a-Service, the company’s flagship offering. This service helps organizations deploy AI quickly. It also boosts operational efficiency and enables customers to uncover new business opportunities.
SambaNova has developed its own chip architecture. Its Cardinal memory is used by the Department of Energy at Lawrence Livermore and Los Alamos. This technology is also deployed in autonomous vehicles and consumer electronics.
SambaNova recently deepened its technical talent and executive leadership team. Its Vice President of Software Engineering is Peter Buckingham, who has been involved in the creation of numerous technologies over the last 20 years.
SambaNova’s Dataflow-as-a-Service delivers the industry’s most powerful integrated AI software-defined hardware platform. The company offers two cloud-like services to enterprise customers. The first is a platform for building advanced AI into internal applications. The second is an API-driven platform that allows enterprises to access SambaNova’s AI system without having to maintain it.
Dataflow-as-a-Service
Using SambaNova’s Dataflow-as-a-Service, you can get state-of-the-art deep learning models and Computer Vision functionality without the hassle of maintaining your own AI system. This software-defined hardware platform delivers foundational models for Natural Language Processing and Computer Vision, plus the company’s Recommender Systems, which are pre-trained models. This helps you to get to market quicker and unlocks new revenue opportunities.
Earlier this year, SambaNova closed a $676 million round of funding. The firm was founded by a group of hardware and software veterans. They have since refocused on selling software-based AI systems to enterprises. This is in response to customers’ desire to focus on a single product or service. The company’s flagship offering is Dataflow-as-a-Service. It’s a nifty little system that uses a proprietary AI chip to power the systems that enable your organization to reap the benefits of AI. It’s a system that can be co-located on your premises or deployed through the cloud.
DataScale
During the second day of the AI Hardware Summit in Palo Alto, California, SambaNova Systems unveiled an updated platform for the company’s DataScale. The company said the new system will be available to the scientific community via the ALCF AI Testbed. The platform is designed for deep learning workloads, which are defined by large models and sparse structures.
The new system features a reconfigurable processor based on SambaNova’s Reconfigurable Dataflow Unit (RDU). The company has not released pricing for the DataScale SN30, but expects to have it ready for sale by late summer. The SN30 is a step up from the older SN10, and SambaNova claims that it can train GPT models faster than any other system in the market. It also has a more advanced memory capacity, and is suited for the largest and most challenging AI models.
Conclusion
The DataScale system has been used by Los Alamos National Laboratory for modeling complex quantum chemistry, and is being coupled into the Corona supercomputing system at Lawrence Livermore National Laboratory. The system has been shown to improve efficiency by allowing developers to think software first.