Speedata, a Tel Aviv-based semiconductor startup developing specialized analytics processing units (APUs) designed specifically for accelerating artificial intelligence and big-data analytics workloads, announced today it has closed a $44 million Series B funding round. The new round brings the company’s total funding to date up to $114 million.
The Series B investment was led by existing backers Walden Catalyst Ventures, 83North, Koch Disruptive Technologies, Pitango First, and Viola Ventures. Strategic investors also joined the round, notably Lip-Bu Tan, CEO of Intel and Managing Partner at Walden Catalyst Ventures, and Eyal Waldman, co-founder and former CEO of Mellanox Technologies.
In contrast to graphics processing units (GPUs), originally created to handle graphical workloads and later adapted to AI and data processing tasks, Speedata says its APUs are engineered from the ground up explicitly to overcome the bottlenecks inherent to big-data analytics. According to company CEO Adi Gelvan, the company’s chip solution can take on workloads traditionally requiring large-scale server clusters fueled by general-purpose processors.
“For decades, data analytics workloads have largely relied upon generic silicon originally meant for other applications,” Gelvan explained. “More recently, GPUs have come into play, propelled by companies like Nvidia specifically for AI-driven processing. Yet these processors, despite their adaptability, are fundamentally not optimized for data analytics. By contrast, our APU technology is purpose-built from inception, enabling a single APU to deliver the processing power of entire racks of conventional servers.”
Established in 2019 by a group of industry veterans, including pioneers in Coarse-Grained Reconfigurable Architecture (CGRA) research, Speedata has focused from the outset on addressing the inefficiencies and energy demands of traditional processors when handling complex analytics workloads. The company says its first-generation APUs are particularly geared toward Apache Spark workloads, with plans to broaden support across other widely used analytics platforms in the near future.
Gelvan stated, “Our ambition is for APUs to become the gold standard for data analytics processing, akin to how GPUs have become synonymous with AI model training.” The company already reports significant interest from major enterprise customers, though it is not yet naming them publicly. Speedata plans to showcase its technology officially at the upcoming Databricks Data & AI Summit, slated to begin in mid-June.
In a notable performance demonstration, Speedata claims its new APU completed a pharmaceutical analysis workload in just 19 minutes—a massive 280-fold improvement compared to the 90 hours typically required using conventional processors.
The completion of the Series B round follows several critical milestones since Speedata’s last funding raise. Last year, the firm wrapped up the design and manufacturing phases for its first-generation chip, moving beyond FPGA-based proof-of-concepts and into fully operational, fabricated hardware. Gelvan noted that the startup is now gearing up for rapid scaling of its commercial activities, given its pipeline of eager enterprise adopters.
“We believe our decades of cumulative silicon innovation will prove transformative in redefining how data-centric workloads are processed throughout the industry,” Gelvan said.
