Marshall Choy, VP Product, and Director Vijay Tatkar outline SambaNova’s sharp growth, Dataflow-as-a-service offering and broad range of AI opportunities
It’s been quite a journey for SambaNova Systems since it was founded by a group of industry luminaries and tech experts in Palo Alto, California, in November 2017. From humble beginnings at Stanford University, the company has grown to employ over 400 staff worldwide, which should increase by another 100 by the end of this year.
SambaNova secured $678 million in Series D funding, led by Softbank Group’s Vision Fund 2, in April, and has already picked up several prestigious awards, including recognition as a Gartner “Cool Vendor”, “Best AI Product for Next Generation Infrastructure” from CogX, and the VentureBeat “AI Innovation for Edge AI” award, among others.
Marshall Choy and Vijay Tatkar – respectively Vice President of Product and Director, Product and Partner Engagements– talk through the company’s meteoric rise and broad potential of industry opportunities.
Choy explains how two of its co-founders, Professor Kunle Olukotun and Chris Ré, were serving at EE and CS departments, working on AI and ML algorithms and techniques, domain-specific languages, compilers and run-time technologies.
“They were specifically preparing for this transition of computing we’re going through right now – some refer to it as ‘Software 2.0’, and transition from the old world of transactional processing to AI computing, from ERP to deterministic written software,” said Choy, who worked previously at Sun Microsystems and Oracle, together with Tatkar.
“It became clear there was a need for a different type of infrastructure that provided greater flexibility and performance, and so SambaNova purpose-built a full stack of hardware and software, to run AI and ML workloads more effectively than conventional solutions.”
It took a couple of years to develop and prioritize the advanced research and implement it into an industrialized, enterprise-ready solution. Today, it has been in market for around a year, with product shipping to revenue customers across multiple industries and continents. SambaNova recently accelerated its go-to-market and category creation strategy with the appointment of its first Chief Marketing Officer, Amy D. Love.
“Our goals for the remainder of the calendar year are extending the leadership position in AI and ML, and specifically putting a great deal of investment and resource to grow our presence globally, beyond North America,” Choy said. “As well as expanding our customer base, and efforts around market awareness, we continue to invest heavily in growing our R&D capabilities to deliver world-class products.”
Tatkar believes the AI domain is now bigger than the internet in terms of revenue and cultural aspects. “I dabbled in AI before, in the 80s and I’m super excited to get back – it now has an unstoppable momentum,” he said.
“I came in with a developers’ tools background and have seen architectural inflection points – how the industry changed from CISC to RISC and from single core to multi-core, and now I believe we are at a new point, from general purpose architectures to DSAs, and compilers lead the way in terms of architecture and how the applications are going to be seen. I believe in ‘own the developer, own the market’; this breakthrough is also because of developers researching and developing incredible models”
In the often-complex world of AI, SambaNova’s model is built on simplicity. “The world around us is innovating like crazy, and our job is to ensure that there is easy adoption, and to enable acceleration while breaking existing barriers,” added Tatkar.
SambaNova’s flagship Dataflow-as-a-Service offering, an extensible AI services platform, enables organisations to jump-start AI initiatives overnight by augmenting existing capabilities and staffing, revolutionising accessibility and empowering organisations in every industry to unleash AI’s vast potential with a simple subscription.
The platform is powered by DataScale, an integrated software and hardware platform delivering unrivaled performance, accuracy, scale, and ease of use built on SambaNova’s Reconfigurable Dataflow Architecture (RDA).
Choy said the market is fairly well bifurcated between leading edge innovators, fast followers, and even laggards – one of the real differences are levels of resource.
Leading innovators in the Fortune 20 companies, who have a great deal of financial backing and tend to be design shops in and of themselves; then everybody else is really looking more towards more complete solutions.
“As a result, we’ve come out with a set of different products and services,” added Choy. “For the Fortune 20s, we have DataScale, which is a complete platform for innovation. The interface point for that system, at a developer level around writing python code, and integrating into open source ML frameworks, enables people to experiment with their own models. That’s a great starting point, because it allows them to focus on their own domains of expertise, rather than spending resources on optimising for a hardware platform
“But there’s a broader set of folks who don’t have 3,000 data scientists – they have three, and may grow to six. So for them, we raise the level of abstraction of the stack to the highest levels at the application layer, so that their interface and our technology stack is merely making API calls from their application, and abstracting all the complexities of model development – everything else becomes seamless and invisible to the user, and they simply interact with SambaNova as an ML services provider.”
He said it has to satisfy the needs of different types of customers so they can deploy it at enterprise scale.
“The market is very mature on the enterprise side, but it’s still developing on the AI side, and that’s the exciting part, growing that ecosystem to all our customers,” said Choy, who added it will be focusing on a number of industry verticals, from manufacturing and life sciences to energy, oil and gas, financial services, and the public sector.
“The energy industry is facing extreme pressure to modernise and become more operationally efficient – and that’s going to require transformational technologies, not incremental, so AI absolutely brings that potential into the lens for the energy sector.
“We have a lot of work in oil & gas and renewables sectors,” Choy said. “There are a number of great opportunities across the sector. If you look at the energy sector, there is a forward-thinking mindset, and some of the facilities are among the most advanced IoT facilities in the world, so there is a strong fit with AI.”
“We have noticed a lot of operators wanting something that’s more specific to the assets they own,” Choy continues. “They have this conundrum of wanting to be efficient but often ‘doing it themselves’ – there is an appetite for more operator-specific solutions. However, we’re in the early days of AI, and adoption depends on the digital maturity of organisations.”
Choy sees that the longer term view is that this technology transition in AI computing will be just as big, if not bigger, than the internet transition was a couple of decades ago. AI will be the main change agent for business and technology. He believes this is just the beginning, the likes of which we haven’t seen before, and the impact will be profound. “We are at the cusp of the biggest computing transition that we’ve seen in decades,” he says.