Speedata, a firm based in Tel Aviv that is creating an analytics processing unit (APU) to speed up tasks related to AI and big data analysis, has raised a total of $114 million through a $44 million Series B fundraising round.

Eyal Waldman, co-founder and former CEO of Mellanox Technologies, and Lip-Bu Tan, CEO of Intel and managing partner at Walden Catalyst Ventures, were among the strategic investors who led the Series B round, along with the company’s current investors, Walden Catalyst Ventures, 83North, Koch Disruptive Technologies, Pitango First, and Viola Ventures.

In contrast to graphics processing units (GPUs), which were first created for visuals and then adapted for AI and data-related jobs, the APU architecture concentrates on resolving the particular analytics constraints at the computing level, the business says.

Adi Gelvan, CEO of Speedata, stated that “standard processing units have been used for decades in data analytics, and more recently, companies like Nvidia have invested in pushing GPUs for analytics workloads.” However, these are not chips created specifically for data analytics; rather, they are either general-purpose processors or processors specialized for different tasks. Because our APU is specifically designed for data processing, a single APU can significantly outperform server racks.

Six founders, including some of the first academics to develop Multi-Threaded Coarse-Grained Reconfigurable Architecture (CGRA) technology, established Speedata in 2019. To solve a basic issue, the founders worked with ASIC design specialists: General-purpose processors were handling data analytics. It might be necessary to access hundreds of servers if the workloads become too complicated. In order to complete the operation more quickly and with less energy, the inventors thought they might create a single dedicated processor.

According to Gelvan, “we saw this as an opportunity to put our decades of silicon research into transforming how the industry processes data.”

According to the CEO of the company, its APU presently targets Apache Spark workloads, but it plans to support all major data analytics platforms.

In the same way that GPUs were made the standard processor for AI training, Gelvan told “we want APUs to be the standard processor for data analytics across every database and analytics platform.”

Although it would not identify them, the firm says to have several big businesses testing their APU. In the second week of June, Databricks’ Data & AI Summit is scheduled to host the official product launch. According to Gelvan, the event will be the first time they publicly display their APU.

In a particular instance, according to Speedata, its APU finished a pharmaceutical assignment in 19 minutes, a 280x speedup over the 90 hours it required when utilizing a non-specialized processing unit.

Since its last round of funding, the business declared to have completed the design and production of its first APU in late 2024, among other milestones.

“We’ve moved from concept to testing on a field-programmable gate array (FPGA), and now we are proud to say we have working hardware that we are currently launching. We already have a growing pipeline of enterprise customers eagerly waiting for this technology and we’re ready to scale our go-to-market operations,” Gelvan said.