Cube semantic layer eases data access from Power BI, Excel
New integrations aim to better enable Microsoft users to directly access data in cloud data warehouses, which has been problematic given how different query languages interact.
Cube unveiled an expanded partnership with Microsoft, launching new integrations on Tuesday between its semantic layer and both Power BI and Excel that enable joint customers to better access data for analysis.
Based in San Francisco, Cube is a 2019 startup whose Cube Cloud platform is a semantic layer designed to enable users to eliminate isolated data, establish consistent models and governance, simplify access and exploration, discover data for reuse and easily integrate with APIs.
Data Analysis Expressions (DAX) API for Power BI integrates Cube's semantic layer with Power BI so that joint customers can access live data in cloud data warehouses directly from Power BI using DAX, which is Power BI's native query language.
Cube Cloud for Excel Add-in, meanwhile, uses Cube's Multidimensional Expressions (MDX) API to connect governed data with Excel so users can update spreadsheets with a single click to analyze current data.
The ability to query cloud data warehouses with DAX really is a breakthrough for teams who have struggled to make Power BI work with their preferred data warehouse platforms.
Donald Farmer Founder and principal, TreeHive Strategy
Accessing live data in data warehouses has been a struggle for many Power BI users, according to Donald Farmer, founder and principal of TreeHive Strategy. As a result, Cube's new integration with Power BI and its semantic layer is a significant addition for joint Cube and Microsoft customers.
"The ability to query cloud data warehouses with DAX really is a breakthrough for teams who have struggled to make Power BI work with their preferred data warehouse platforms," Farmer said, noting that even using Power BI with Microsoft's own Fabric platform has been a struggle for data architects and engineers.
Regarding the integration between Cube's semantic layer and Excel, Farmer added that it is also significant, given that it simplifies connections between Excel and cloud data storage platforms.
"The Excel connectivity is also an excellent addition," he said.
Cube and Microsoft first partnered in March 2024.
New capabilities
Data architects and engineers have long been able to connect Power BI and Excel directly to data warehouses such as Azure or third-party platforms including Databricks and Snowflake.
Such direct connections, however, don't always work smoothly.
Microsoft developed MDX in the late 1990s to connect analytics tools with multidimensional online application processing (OLAP) cubes. DAX was developed later by Microsoft as an intended improvement and became the query language for Power BI.
Neither, however, is the current industry standard. Instead, SQL has become the query language for most analytics and data warehouse platforms.
Before Cube's new integrations, Power BI and Excel users had to either copy and move data from data warehouses via import mode or MDX and DAX needed to be translated to SQL in Microsoft's DirectQuery mode. Copying and moving data can be labor intensive, while translations from Microsoft's query languages to SQL are not always seamless, which leads to a lack of performance, according to Artyom Keydunov, Cube's founder and CEO.
The integrations between Cube's semantic layer and Microsoft's analytics platforms are intended to address query performance.
"The SQL generated is frequently unoptimized and performs poorly," Keydunov said. "Our aim with the [the integrations] is to improve this performance."
Given that the integrations improve query performance between Power BI and Excel and cloud data warehouses, they address two major trends, according to Kevin Petrie, an analyst at BARC U.S.
One is the sustained popularity of spreadsheets. The other is that data remains highly distributed despite the efforts of cloud data platforms to help organizations consolidate. As a result, the integrations are significant.
"This announcement gives companies a useful method of analyzing data," Petrie said. "Analysts and data teams of all types need to access distributed data wherever it resides in order to drive decisions and support increasingly advanced models."
Like Keydunov, Farmer noted that direct connections between Microsoft's analytics platforms and data warehouses often have poor results. In addition, for those choosing to use DirectQuery, costs can add up, he continued.
Cube's semantic layer enables users to cache data, which results in more efficient access from Power BI and Excel. And it enables such access via DAX and MDX.
"By enabling connectivity with both these standards, Cube has created a modern OLAP solution, which is a real breakthrough in engineering for them," Farmer said.
Regarding the impetus for integrating Cube's semantic layer with Power BI and Excel, customer feedback was a significant factor, according to Keydunov.
Power BI is perhaps the most widely used business intelligence platform, with more than 12 million users. Excel, meanwhile, remains the most popular tool for business analysis with more than 750 million users.
"The continued investments in new Microsoft integrations are a direct response to enterprise customer demand for these capabilities," Keydunov said.
Future plans
In addition to its partnership with Microsoft, Cube is partners with AWS, Google Cloud, Databricks and Snowflake.
However, despite integrations with prominent data platform vendors, Cube's semantic layer platform is a relative newcomer compared with others providing similar capabilities such as AtScale, GoodData, Looker and MicroStrategy. In addition, its total funding of $46.7 million, including $25 million in 2024, is far less than other competitors such as DBT Labs.
To compete, one of Cube's goals is to continue modernizing OLAP, according to Keydunov. Another is to emerge as a catalyst for AI adoption by enabling customers to turn raw cloud data into AI-ready data.
"With well-defined semantic modeling, it becomes possible to deliver consistent, reliable and trustworthy AI outputs and autonomous actions," Keydunov said.
That focus on supporting AI platforms is wise, according to Petrie.
Universal semantic layers are a valuable way to unify access to distributed data, he noted. Migrating data, data sovereignty requirements and security risks often prevents organizations from consolidating data in one location. Semantic layers help address that sprawl.
Access to data, meanwhile, is critical for AI development. Cube now supports such platforms as the LangChain framework. However, there are opportunities to integrate with others, according to Petrie.
"I would recommend they consider extending their support to include other AI/ML platforms, such as Dataiku and Domino Data Lab," he said. "Data scientists need easy access to structured data as they train advanced models and put them into production."
Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than 25 years of experience. He covers analytics and data management.
