AI on Demand: SambaNova Hopes to Bridge the AI Integration Gap

The artificial intelligence (AI) and machine learning (ML) trend has been booming recently. From startups to established companies like NVIDIA, AI is a crucial focal point for new up-and-coming technology. Recently, a new investment program was started called "OpenAI Startup Fund," which aims to give startups funds to make a "profound, positive impact on the world."

 

AI deployment and exploring rates for a few key countries. Image used courtesy of IBM

 

This focus makes sense with the intended benefits of utilizing and creating functional AI. On an individual level, it brings consumers better experiences and could help them make more informed decisions. On a commercial level, it allows businesses to process and understand their data more efficiently, and it is widely seen as a key stepping stone into the future. 

While AI tends to play a more passive role on an individual level, businesses, particularly those that are not tech firms at heart, might have to do a lot of heavy lifting and shifting to extract the maximum benefit from it as a technology. This integration can be expensive, time-consuming, and problematic if the AI systems aren’t implemented correctly, making it all the more important to make AI systems easier to incorporate.

Now, a U.S. startup, SambaNova Systems, aims to bridge the gap between businesses and AI, and it has recently been awarded a considerable amount of funding to do just that. 

 

Bridging the Gap

SambaNova Systems, a California-based chipmaker, builds AI hardware and the integrated systems that run on it. While the company was founded in 2017, it only officially came to market in December 2020. 

This company was co-founded by two Stanford University professors—Kunle Olukotun and Chris Ré—and Rodrigo Liang, an engineering executive at Oracle. With these leaders at the helm, SambaNova has raised more than $1 billion to date to build out its AI-focused hardware known as 'DataScale' and the system that runs on it, 'Samba.' 

 

DataScale Overview

DataScale is an integrated software and hardware systems platform. According to SambaNova, it uses a software-defined-hardware approach that delivers "unmatched efficiency across applications." 

At the core of DataScale is SambaNova's 'Reconfigurable Dataflow Unit' (RDU), a next-generation processor designed for dataflow-orientated workloads such as machine learning and high-performance computing. 

 

RDU dataflow overview. Image used courtesy of SambaNova

 

SambaNova's business means that, on one level, it competes with the likes of Nvidia and other startups such as Graphcore. However, the company has a unique approach to tackling AI challenges for businesses.

 

Dataflow-as-a-Service

In December, the company launched 'Dataflow-as-a-Service,' a subscription-based service that allows businesses to utilize SambaNova's AI system on-demand. 

 

SambaNova's Data-as-a-Service ML platform. Image used courtesy of SambaNova

 

The service follows CEO Liang's belief that there is significant value in selling software-based AI systems to enterprises that want to integrate more AI into their operations but might lack the knowledge and other resources to be successful.

"Rather than you coming in and thinking about how to hire scientists and hire and then deploy an AI service, you can now subscribe, and bring in that technology overnight. We're very proud that our technology is pushing the envelope on cases in the industry," Liang stated in a press release. 

In April, SambaNova raised $676 million in a Series D funding round which will provide key funds to further its goals for easier AI integration. 

 

Stirring Up Industry Excitement

SambaNova's DataScale platform is just one of many recent developments that together represent a new wave of semiconductor innovations in the U.S. market. 

There is starting to be an uptick in the general enthusiasm for chip startups born from necessity with the chip shortages. Also, more and more businesses seem to be interested in purpose-built processors that can efficiently run AI technologies and machine learning algorithms tasked with wading through vast troves of data. 

Startups like SambaNova, Cerebras Systems, Groq, and others make chips that claim to perform certain AI tasks better than those from other big names like Nvidia, who currently dominate the market with their graphics processing units (GPUs). 

With this influx of interest, investment, and need, this trend of AI-based hardware will be sure to produce a significant amount of new and inspiring technology.