速递｜B轮融资4400万美元，以色列创企Speedata获英特尔CEO站台，APU芯片让数据分析告别“服务器农场”

Core Insights - Speedata, a startup based in Tel Aviv, has developed an Analysis Processing Unit (APU) designed to accelerate big data analytics and AI workloads, completing a $44 million Series B funding round, bringing its total funding to $114 million [2][3] - The APU architecture focuses on addressing specific bottlenecks in computational analysis tasks, differentiating itself from GPUs, which were originally designed for graphics processing [2][3] - Speedata aims for its APU to become the standard processor for data processing, similar to how GPUs have become the standard for AI training [4] Funding and Investment - The Series B funding round was led by existing investors including Walden Catalyst Ventures, 83North, Koch Disruptive Technologies, Pitango First, and Viola Ventures, along with strategic investors like Intel's CEO Lip-Bu Tan and Mellanox Technologies' co-founder Eyal Waldman [2] - Since its last funding round, the company has achieved several milestones, including the design and manufacturing of its first APU by the end of 2024 [4] Product Development and Performance - The APU is currently focused on Apache Spark workloads, with plans to support all major data analytics platforms in the future [3] - Speedata claims that its APU can complete a pharmaceutical workload in just 19 minutes, compared to 90 hours using non-specialized processors, achieving a performance improvement of 280 times [4] - The product is set to be officially launched at Databricks' Data & AI summit in the second week of June [4] Company Background - Speedata was founded in 2019 by six co-founders, some of whom were early researchers in multi-threaded coarse-grained reconfigurable architecture (CGRA) technology [3] - The founding team collaborated with ASIC design experts to address the fundamental issue of data analysis workloads being handled by general-purpose processors [3]