Cube is building a ‘semantic layer’ for enterprise data

An increasing number of companies are adopting data models: abstract models that organize data elements and standardize how they relate to each other. But as the rise of data analytics and AI leads organizations to expect more from data models, many of the old paradigms are proving unwieldy and exceptionally fragile.

At least that’s what engineers and businessmen Artyom Keydunov and Pavel Tiunov observed in their work. While at Starsbot, a data analytics startup the pair co-founded in 2016, Keydunov and Tiunov often consulted with organizations struggling to get their “data house” in order, so to speak.

Cube began as an open source project in 2019 offering what Keydunov describes as a “universal semantic layer” for organizational data that can feed databases, business intelligence (BI) tools, and even AI-powered chatbots. Now, five years later, Keydunov and Tiunov have a real business on their hands, having launched a subscription-based service built on Cube (Cube Cloud) that adds automated workflows and enterprise-focused governance and deployment tools.

“There is no missing data,” Keydunov told TechCrunch. “And demand for data continues to grow among employees, partners and customers, who are motivated by the idea that data-driven decisions lead to greater operational efficiency, greater customer satisfaction and competitive advantage. Technologies such as AI, machine learning, the Internet of Things and blockchain are reshaping the data landscape and revolutionizing the way organizations collect, process and derive value from data. It’s not just humans who need data; Now machines also need data.”

Data modeling challenges aside, surveys suggest that relatively few organizations are achieving even basic-level success in deriving value from their data. A 2022 Gartner survey of data analytics leaders found that less than half believe their teams are effective at delivering value to their employers. This despite the fact that, according to the same survey, companies are spending an average of more than $5 million on data management, governance and analytics initiatives.

So what to do? For Keydunov and Tiunov, the answer was to try to create a platform that could serve as a unified source of truth for all of a company’s data and metrics.

An illustration of Cube’s semantic data layer.
Image credits: Cube

“Cube Cloud is a universal semantic layer that is an independent, but interoperable, part of the modern data stack that sits between data sources and data consumers,” Keydunov said. “The universal semantic layer allows all data endpoints (whether BI tools, embedded analytics, or AI agents and chatbots) to work with the same semantics and underlying data.”

Enterprises use Cube Cloud to build that semantic layer and connect it to their various applications and utilities, employing role-based access controls, data caching, single sign-on, and scaling infrastructure as required. Enterprise-level customers get access to consultants who can train their data engineers to work with Cube Cloud and offer on-demand support, as well as create the initial Cube Cloud instance, either on Cube-owned servers or on-premises, personalized for the business.

“Cube Cloud automatically adjusts queries and injects the appropriate security context (user or role details) to ensure that only the appropriate users have access,” adds Keydunov. “And through Cube’s performance insights, customers can find redundant queries or other opportunities to cache and pre-aggregate query results, reducing the amount of processing required.”

Cube competes with AtScale, which also offers a semantic layer for data modeling and serving, and Transform, recently acquired by Dtb Labs. But Cube appears to be holding its own, with a customer base spanning more than 200 Fortune 1000 brands and a of users that is close to 5 million people, says the company.

Keydunov claims that the Cube open source project has surpassed 10 million downloads, while Cube Cloud is currently installed on about 90,000 servers. Bookings tripled between 2023 and 2024, while the average deal size tripled.

It is undoubtedly this success that attracted new investment into the business. San Francisco-based Cube announced this week that it raised $25 million in funding from backers including Databricks Ventures, Decibel, Bain Capital Ventures, Eniac Ventures and 645 Ventures. Bringing the total raised by the 40-employee startup to $48 million, the new cash will go toward supporting Cube’s marketing and merchandising activities and expanding Cube Cloud’s capabilities, Keydunov said.

“Our investors encouraged us to raise capital to support the expansion of our go-to-market team so we could take advantage of the huge increase in demand for AI and the semantic layer,” Keydunov continued. “We have seen companies become more measured and careful in their evaluations, which may slow down the sales process a bit, but that gives us more time to demonstrate our value over the competition. “We are well capitalized with our new funding round and have a long way to go to grow the company to its next milestone.”

Leave a Comment