Qbeast gets $7.6M in funding to streamline data lakehouses queries with multidimensional indexing

Data optimization startup Qbeast Analytics Inc. says it’s going to help organizations eliminate the “hidden tax” of big data lakehouse frameworks like Delta Lake after raising $7.6 million in seed funding today.

The round was led by Peak XV’s Surge, formerly known as Sequoia Capital India. Other investors were HWK Tech Investment and Elaia Partners. The money will be used to fund new hires and expand its platform to new use cases, the company said.

According to Qbeat, open-source lakehouse architectures such as Delta Lake, Apache Iceberg and Apache Hudi have become extremely popular in the enterprise today, helping organizations to keep a lid on exploding data volumes that strain traditional infrastructures. However, although these platforms are incredibly powerful and highly scalable, Qbeast points to an enormous “tax” on using them, with up to 90% of compute resources being wasted on scanning irrelevant data stored within them.

The problem is that though data lakes are massive, they’re not smart about the way they organize and search through that information. As a result, queries are becoming both painfully slow and very expensive. “There is an undesirable compute cost hidden in the data layout that has been highly neglected by the market for data lakehouses,” said Chief Technology Officer Flavio Junequeira, who previously helped create Apache BookKeeper and Apache ZooKeeper.

Qbeast, which was born as a research project at the Barcelona Supercomputing Center, has developed the solution – a data optimization platform that plugs directly into existing Delta, Iceberg and Hudi tables. It features multidimensional indexing capabilities with complex filters across columns such as time, region and customer segment, enabling customers to search through and query only the data they need to look at.

The platform can also optimize for either historical or real-time queries in a single table, with simultaneous filtering across any mix of data attributes, in contrast to traditional partitioning tools that only work in single dimensions. What’s more, Qbeast’s drop-in indexing works perfectly with compute engines such as Databricks, DuckDB, Polars, Snowflake and Spark, so there’s no need to rewrite data pipelines or mess with the underlying storage layer. By making data more efficient to query, Qbeast says, it can increase query speeds by up to six times, depending on the dataset, while reducing compute costs by as much as 70%.

These are some impressive claims, and that’s precisely what convinced former Microsoft Azure and Amazon Web Services cloud infrastructure veteran Srikanth Satya to accept the role of the company’s new chief executive.

Satya, whose appointment was also announced today, said data teams shouldn’t be forced to choose between speed, cost or openness. “We built Qbeast to make high-performance analytics simple and accessible, without locking organizations into proprietary systems,” he explained. “In a world where data is growing faster than ever, we’re here to ensure every company can turn that data into value on their own terms.”

Another advantage of Qbeast is that its platform plays natively with existing data tools, being compatible with all of the most popular open data formats, so there’s no need for customers to make any changes to their data infrastructure at all. They can just drop in its multi-indexing tool and leave everything else as it is.

“We believe every organization, not just the tech elite, should be able to extract value from their data without incurring massive cloud costs or hiring a team of engineers to tune performance,” Satya added.

Michael Ni of Constellation Research Inc. said Qbeast’s funding round signifies a broader trend in the storage industry, as new and old vendors alike look to figure out ways of fixing bottlenecks in scaling AI-based insights and decision-making. He explained that the push for AI-driven automation means that business advantage is no longer determined by tools, but the platforms that can eliminate friction and boost performance.

“Even as data lakehouses become the common data foundation for enterprises, they are growing with behavioral and multimodal data for AI agents,” Ni said. “But these AI agents can’t operate in real-time if your platform still relies on partitions and full scans, and that’s why Qbeast is targeting the pressure point with fast, flexible and multi-attribute data access, without vendor lock-in or cloud dependence.”

Though Qbeast’s basic platform is up and running and available now, it has an extensive roadmap worked out that will see it add new capabilities including auto-tuning, adaptive indexing and deeper data engine support, across cloud infrastructure providers and industry verticals. Ultimately, it wants to become the default indexing layer for every open lakehouse architecture, paving the way for a future where data innovation isn’t hindered by spiraling compute costs or performance constraints.

HWK TechInvestment CEO and Managing Partner Juan Santamaría said Qbeast is solving an important and fundamental problem. “Its multi-dimensional indexing layer has the potential to become critical for every company moving to a lakehouse model,” he said.

Images: Qbeast Analytics