Speedata, a Tel Aviv-based startup, has raised US$44 million in series B funding to develop analytics processing units (APUs) for big data and AI.

The round was backed by existing investors like Walden Catalyst Ventures, 83North, Koch Disruptive Technologies, Pitango First, and Viola Ventures, as well as strategic investors such as Intel CEO Lip-Bu Tan and Mellanox co-founder Eyal Waldman.

Founded in 2019, Speedata designs APUs tailored for data analytics to reduce processing bottlenecks, aiming to outperform GPUs originally built for graphics.

The startup claims its APUs can replace large server setups, offering better performance and energy efficiency, and will debut the chip at the Databricks Data & AI Summit in June.

In testing, Speedata’s APU completed a pharmaceutical workload in 19 minutes versus 90 hours with standard processors, and the company is now preparing to scale.

🔗 Source: TechCrunch

🧠 Food for thought

1️⃣ Specialized processors are reshaping big data economics beyond AI

Speedata’s APU represents a growing trend of specialized chips targeting specific computing workloads outside of AI training, where Nvidia’s GPUs have dominated.

The startup’s approach tackles a fundamental market gap: while GPUs were adapted from graphics processing for AI, they weren’t specifically designed for database and analytics workloads that have different computational requirements 1.

Their claims of 20x to 100x performance improvements over traditional CPUs for analytics tasks suggest potential for dramatic infrastructure consolidation, where a single APU could replace multiple racks of servers 1.

This follows a historical pattern in computing where general-purpose processors eventually give way to specialized hardware for efficiency, such as how TPUs emerged for specific AI workloads.

The $44 million Series B round following their earlier $70 million in funding demonstrates strong investor confidence in purpose-built analytics processors, suggesting the market sees substantial value in workload-specific acceleration beyond AI 1.

2️⃣ Industry-specific acceleration signals new frontier in computational research

Speedata’s pharmaceutical industry results—completing a compound similarity analysis in 19 minutes versus 90 hours on traditional processors—demonstrates how purpose-built hardware can fundamentally transform research timelines in data-intensive fields 2.

The 280x performance improvement in pharmaceutical applications could significantly accelerate drug discovery processes, potentially allowing researchers to explore vastly more chemical compounds in the same timeframe 2.