The growing demand for generative AI apps is driving larger and larger databases to store the relevant data (e.g., model training data). These databases tend to be resource intensive from a hardware perspective and can have high latency depending on the algorithms used to orchestrate them. Companies are often forced to make trade-offs between database cost, performance, and accuracy.
But it doesn't have to be that way, says Ohad Levi, CEO and co-founder of Hyperspace, which builds custom cloud instances to speed up two specific database tasks: lexical search and vector search. Lexical search is a type of keyword-based search that looks for an exact match in a database, while vector search takes into account the semantic meaning and context of the search query.
Levi claims that Hyperspace instances that leverage a combination of FPGAs and GPUs can deliver searches up to 10 times faster than traditional, non-accelerated databases.
“Our products help companies deal with large-scale data search, especially in AI and generative AI applications,” Levy told TechCrunch. “Unstructured data is outpacing traditional search capabilities. Data search solutions need to accommodate lexical and vector search datasets to meet current market demands.”
Prior to starting Hyperspace, Levi was an optimization engineer at Intel and then a product marketing leader at HP. He says he was frustrated with the limitations of traditional search solutions for large technology companies and partnered with former Intel design consultant Max Nigiri to found Hyperspace.
Hyperspace doesn't sell instances, but instead sells access to managed database software (currently hosted by AWS) that runs on those instances. Hyperspace's databases can handle many kinds of structured and unstructured data, including video, images, and text, and are priced according to size and query volume.
“Hyperspace operates as a software-as-a-service model with pay-per-use and is a cloud-native, managed database,” Levi explains. “Our team can design customized AI infrastructure solutions to help companies solve their search challenges.”
Hyperspace's performance gains, if real, are impressive: Levi claims that its instances deliver five times the throughput at 50% lower cost than typical databases. (These are average results, and Levi declined to compare directly with competitors.) But with so many incumbent platforms to choose from, including Azure, AWS, and Google Cloud, can Hyperspace convince enterprises to use a newbie database platform?
Levy says yes, claiming that Hyperspace has already seen success with early customers. The Tel Aviv-based company has tripled its annual recurring revenue and total contract volume over the past year, signing companies in the fraud prevention and e-commerce sectors, including Forter, Nsure and Renovai.
Hyperspace also recently closed a $9.5 million seed funding round led by Mizmaa, with participation from JVP and toDay Ventures. Levi said the funding will be used to expand Hyperspace's database offering to “thousands” of instances and launch a free entry-level plan.
“Hyperspace has a pipeline of innovative new products that will drive the search market forward and serve the needs of our enterprise and small business customers,” said Levi. “We're not seeing any headwinds. Every generative AI system is a search system, and search is harder than it used to be. The need for better AI infrastructure grows every day, and as we have more data, the need for better search applications becomes more and more apparent.”