NeuroBlade
Definition
NeuroBlade is an Israeli artificial intelligence company founded in 2018. It develops dedicated hardware to speed up the processing and analysis of large databases at industrial scale.
The problem: fast analysis of huge databases
Large organizations run queries on databases and data warehouses of enormous size, and the general-purpose computing model based on a processor (CPU) becomes more expensive and slower as the data grows. NeuroBlade, founded in 2018 by Elad Sity (CEO) and Eliad Hillel (CTO), developed a hardware answer: a dedicated data analytics chip called the SQL Processing Unit (SPU). It works with common processing frameworks such as Spark, Presto and ClickHouse and, according to the company, speeds up queries 4 to 10 times compared with leading processors.
The company raised more than $100 million from strategic and financial backers, including the Taiwanese chip giant MediaTek and Intel, and operated from Tel Aviv and Mountain View, California. The Israeli company Speedata works in the same field, developing a dedicated processor for analytical processing (APU). According to the company, one server with its processor can replace several racks of regular processors.
The Amazon partnership: accelerated Amazon EC2 F2 instances
Before the acquisition, NeuroBlade worked with Amazon on accelerated instances in the Amazon EC2 F2 cloud service, which is based on AMD FPGA chips. NeuroBlade's SQL processor (SPU) runs on these chips and speeds up large-scale data processing. It is an example of how a small startup can deliver value to a giant provider's cloud infrastructure without building data centers itself.
In 2023 NeuroBlade won the 'Most Innovative AI & Big Data Analytics Application' award at the Flash Memory Summit, industry recognition of its dedicated-hardware approach, relatively shortly before the next major step in its history.
The end: Amazon takes over the engineering team
In October 2025, NeuroBlade announced that it had signed a binding agreement to transfer its core engineering team to AWS Annapurna Labs, Amazon's chip development arm in Israel, which is responsible among other things for the Trainium training processors used to run AI models in the cloud. An announcement of the deal still appears as the headline on NeuroBlade's own official website.
The deal, estimated at tens of millions of dollars, is seen in the industry as the effective end of NeuroBlade's independent operations: not a full company acquisition, but an acqui-hire of the development team. The company's CEO, Elad Sity, wrote to employees that this was the next chapter for the team, which would continue developing the technology on a larger scale inside Amazon. The acquisition illustrates a broader trend in the data hardware industry: small, independent players find it hard to compete with the huge development budgets of the cloud giants.
FAQ ❓
What did NeuroBlade do?
It developed a dedicated chip (SPU) to speed up queries and analysis on huge databases, as a faster and cheaper alternative to analysis based on general-purpose processors.
What happened to NeuroBlade in October 2025?
Its core engineering team moved to AWS Annapurna Labs, Amazon's chip arm in Israel, according to the company's own announcement. The deal is seen as the end of its independent operations.
How much money did NeuroBlade raise?
More than $100 million in total, from investors such as MediaTek, Intel, Grove Ventures and Maverick Capital, before its engineering team moved to Amazon.
When was NeuroBlade founded?
In 2018, by Elad Sity and Eliad Hillel, with operations in Tel Aviv and Mountain View, California.
Found a mistake, or have a source we should add? Write to us