EnterpriseDB on Tuesday announced new EDB Postgres AI features for WarehousePG, including per-core pricing to help users control AI development costs and streaming data capabilities aimed at feeding real-time data to AI applications.
Announced in April, WarehousePG is an open source PostgreSQL data warehouse based on source code from the Greenplum Database project. WarehousePG is part of EDB Postgres AI, a database platform that unifies data management, analytics, and AI, introduced in May 2024.
PostgreSQL is a relational database system known for its flexibility and versatility. Beyond traditional relational database capabilities, PostgreSQL supports geospatial, JSON, time series, and vector database workloads. A PostgreSQL data warehouse serves as a central repository for data that feeds analytics and AI workloads.
WarehousePG's new features include streaming data and a predictable pricing model, as well as upgraded data observability and data sovereignty capabilities, and flexible deployment across any cloud, on-premises, and region.
Carl Olofson, an analyst who recently founded DBMSGuru after 28 years at IDC, said the WarehousePG update is a valuable addition because it provides EnterpriseDB users with some of the functionality they need for AI development.
“From an AI perspective, the real-time streaming offering and observability enhancements are important updates because the resulting package will compete with other leading commercial data warehouse relational database products for AI,” he said. “Focusing on sovereignty is also important, as it addresses key issues when applying AI to data that has international sources.”
Wilmington, Delaware-based EnterpriseDB competes with database vendors such as MongoDB and MariaDB, as well as hyperscalers offering PostgreSQL databases such as AWS, Google, Microsoft, and Oracle.
WarehousePG upgrade
Advances in AI technology, starting with OpenAI's launch of ChatGPT in November 2022, have led many companies to increase their investments in AI initiatives.
From an AI perspective, the provision of real-time streaming and enhanced observability are important updates as they make the resulting package competitive with other leading commercial data warehouse relational database products for AI.
Karl OlofssonFounder, DBMSGuru
Generative AI (GenAI) Applications such as chatbots and AI agents simplify data exploration and analysis, empowering employees to make informed decisions. It also helps organizations automate certain tasks and processes and improve overall efficiency.
However, building chatbots and agents can be expensive.
To perform well, chatbots and agents require much more high-quality data than is required to inform traditional analytical tools such as reports and dashboards. Therefore, ingesting, integrating, preparing, and monitoring data for AI takes more time than traditional analytics. They also require much more computing power than analytics tools, both for storage and for running workloads.
As a result, companies spend far more money developing AI tools than building analytical applications. To help customers manage AI-related spending, many vendors are focusing on cost management.
For example, AWS aimed to control costs with data management features announced at the re:Invent user conference in early December. Similarly, database vendors such as Aerospike and Neo4j have added performance improvements aimed at helping customers save money.
EnterpriseDB aims to control costs by charging WarehousePG users with a per-core pricing model rather than pay-as-you-go pricing.
Per-core pricing fixes costs by charging customers based on the number of CPU cores used in their WarehousePG deployment. The pay-as-you-go pricing model fluctuates based on your usage, so it can cost you a lot more than you expected.
According to BARC US analyst Kevin Petrie, EnterpriseDB's per-core pricing for WarehousePG is important given that the cost of AI development is a concern for many enterprises.
“AI adopters will appreciate per-core pricing as an alternative to pay-as-you-go,” he said. “BARC research shows that software is the biggest cause of cost overruns in AI projects, and this reduces that risk.”
In addition to per-core pricing, the data streaming capabilities that EnterpriseDB has added to WarehousePG are also valuable, Petrie continued.
Stream processing is the continuous, high-speed movement of data from a source to an application through a data streaming platform and is a key feature of agent AI pipelines that enables agents to act on real-time information.
“Given the low latency of modern workloads, we expect to see significant demand for data streaming,” Petrie said. “Many of the most common AI use cases, such as fraud detection, preventive maintenance, and price optimization, can benefit from EnterpriseDB’s data streaming support.”
“Streaming support is essential if you want to incorporate AI processing into your workflows, not just for end-user queries,” he said.
In addition to per-core pricing and new data streaming capabilities, the WarehousePG update for EnterpriseDB includes:
AI-enabled architecture that includes stream processing, native vector search and storage capabilities, and in-database machine learning using Python and MADlib.
Deploy flexibly across regions in any cloud or on-premises.
Data governance features, including data observability to monitor anomalies and changes that can impact AI output.
While predictable pricing and streaming capabilities stand out, WarehousePG's updates were driven by conversations with customers, said Quais Taraki, chief technology officer at EnterpriseDB.
“This update was decidedly customer-driven,” he said.
In addition, EnterpriseDB's market research shows that 95% of enterprises plan to integrate data and AI in the next three years, placing a strong emphasis on data sovereignty, which played a role in this release, Taraki continued.
“These features were prioritized to reflect this reality,” he said.
Looking to the future
With the latest WarehousePG updates now available, EnterpriseDB will focus on improving the interoperability of its EDB Postgres AI platform in 2026, according to Taraki.
Database platforms offer a wide range of functionality. But to best serve customers building AI and analytics applications, integrations that allow easy connectivity with external platforms are critical.
“Deepening interoperability across the AI and analytics ecosystem will be a key focus in 2026,” Taraki said. “This means better business continuity and security, better integration with open data and AI frameworks, and fewer operational barriers for customers modernizing their analytics workloads with Postgres.”
EnterpriseDB's emphasis on interoperability is smart, Olofson said, noting that more partnerships and integrations will help position the vendor's database products within the broader analytics and AI ecosystem.
“EnterpriseDB's platform is very competitive on its own, but if enterprises want to use it in complex data management environments, especially those that support AI, they also need a combination of other technologies,” he said. “The key for EnterpriseDB is partnering with technology suppliers and consultants. [EnterpriseDB PostgreSQL] Services can be brought into the context of business solutions. ”
Eric Avidon is a senior news writer at Informa TechTarget and a journalist with over 25 years of experience. He is responsible for analysis and data management.
