Israeli data accelerator startup NeuroBlade has announced closing a significant $83 million Series B round. The investment was led by Corner Ventures, with participation from existing investors: StageOne Ventures, Grove Ventures and Marius Nacht. Additionally, technology companies including MediaTek, Pegatron, PSMC, UMC and Marubeni also got in on the action.

“Steroids” but for data processing

Today, in almost every field (financial, medical, advertising, etc.), companies have a growing database, which it queries again and again. The ability to solve these queries faster is at the core of growth potential for these organizations, and it is quite a massive market - with many large corporations starting to build dedicated data analytics hardware.

This is where NeuroBlade comes into the picture, the Israeli startup has developed a processing unit that accelerates data analytics and unclogs traditional bottlenecks by integrating its technology into a full system-level easy-to-deploy appliance. “If we break it down into simpler but more technical words, at the basic level we accelerate SQL queries. We built our accelerator based on the development of a new type of memory that allows us to run calculations in the memory itself. The acceleration of SQL is becoming less and less trivial over the years as the amount of data they scan increases, " says CEO and co-founder Elad Sity, in a conversation with Geektime.

He says the idea behind the product came after they recognized that conventional computer systems have difficulty processing large volume information efficiently. So Neuroblade decided to rebuild their architecture with the help of 'computational memory'. “In a classic computer system the memory and the processor are separated from each other and the information has to flow between them. At Neuroblade we have developed a new type of memory where the processing is already done within the memory and there is no need to move the information back to the processor and thus possible to accelerate certain operations 10X, 100X, and even more, " adds Sity.

According to him, Neuroblade aimed to accelerate "memory-driven" AI models, something they do today in server farms around the world. He says that one of the bottlenecks they examined at the beginning in order to understand what can be built and what can be accelerated with the help of computational memory was memory-driven AI models, an area in which they have achieved noteworthy results.

How does your product stand out from the incredible offering from the likes of Intel and NVIDIA? What are your strengths?

"Our XRAM memory is basically a “steroid” for these systems, it allows us to speed up data processing more than any traditional system that uses standard memory. And this, of course, combined with the experience of our team in building end-to-end systems. That is, seeing the big picture and developing all the different parts of the system at the same time so that the value of the whole is greater than the sum of its parts. ”

We asked Sity how the company is coping with the ongoing chip shortage crisis, which is felt heavily by both small manufacturers and market giants alike. According to Sity, the fact that among the company's investors are also some of these giants - like Intel and MediaTek, from China - helps the company a lot. “Our investors also include companies that are part of our supply chain, which reduces our exposure to the crisis. Beyond that, both me and my partner Eliad came from SolarEdge -- a company whose mass production is one of its pillars, so we have experience in managing crises."

NeuroBlade was founded in 2018 by Elad Sity and Eliad Hillel (CTO), graduates of the 81st Technological Unit of the Intelligence Corps and among the first employees at SolarEdge. NeuroBlade currently employs more than 100 people. The capital raised will enable Neuroblade to expand its development and support teams, including sales, support and marketing abroad. To date, the Israeli startup has raised $110 million, after emerging from stealth only two years ago with a $23 million funding round.