Getty Images/iStockphoto
Teradata extends AI development to on-premises environments
For enterprises concerned about cloud cost control and data sovereignty, the longtime vendor's new platform enables users to build advanced tools in their own systems.
Teradata on Tuesday launched AI Factory, a platform that enables users to develop and deploy AI tools on premises.
Teradata provides the same set of capabilities for cloud-based environments through VantageCloud.
Cloud-based AI development has advantages such as managed services that remove the burden of infrastructure management, on-demand computing resources that can scale up or down based on the workload demands of individual projects and flexible access to third-party tools. However, because of the enormous size of AI development workloads, cloud-based development can get expensive. In addition, because cloud providers often store data in numerous geographically located data centers, remaining compliant with regulations is challenging.
Cost concerns and compliance risks related to the cloud make on-premises development, while not as flexible, preferable in certain circumstances. In particular, enterprises in highly regulated industries such as healthcare and government sometimes prefer on-premises environments.
Michael Ni, an analyst at Constellation Research, noted that Teradata's historic strength was as a platform for high-performance analytics in a secure environment. With cloud repatriation a trend, AI Factory plays to that strength.
"AI Factory doesn't invent a new advantage. It reactivates one just as the market is circling back to demand sovereign, secure, high-performance AI on premises," Ni said. "Cloud cost volatility and regulatory scrutiny are rising, prompting enterprises to rethink where their most sensitive and strategic AI workloads run."
David Menninger, an analyst at ISG Research, likewise noted the significance of AI Factory because it addresses data sovereignty concerns.
"Enterprises that have specific concerns about governance, sovereignty and costs will be able to address some of their concerns with AI Factory," he said. "Several of the leading AI platforms are not readily available for on-premises deployments. Teradata's AI Factory [provides] an alternative."
Based in San Diego, Teradata is a longtime data management and analytics vendor. Like many of its peers, which include tech giants such as AWS and Google, Teradata expanded into AI development after OpenAI's November 2022 launch of ChatGPT sparked surging enterprise interest in using AI to make employees better informed and more efficient.
A platform for on prem
With its combination of data management and AI development capabilities, Teradata essentially provides customers with the materials for building applications on a foundation of trusted data, according to Louis Landry, the vendor's chief technology officer.
AI Factory extends those materials to all Teradata users.
"Costs can get crazy [in the cloud]," Landry said. "AI workloads, which can get incredibly compute-intensive, and experimentation are big multipliers. We think [AI Factory is] valuable for anyone looking for data sovereignty and cost control."
In practice, many enterprises have hybrid data environments. While AI Factory now provides Teradata users with the option to develop AI tools on premises, the platform does not limit the use of those tools to on-premises environments.
AI tools built with AI Factory can operate -- and interoperate with other AI tools -- however the enterprise chooses whether in the cloud, on premises or a hybrid environment, according to Landry.
"Our platform is the same, be it on cloud or on premises," he said. "There are a lot of use cases for AI. You have to be able to span across [all environments]."
AI Factory features include the following:
- An infrastructure featuring connectivity to Teradata's ClearScape Analytics, developer tools such as JupyterHub and Airflow, support for model lifecycle management and compliance, and one-click large language model deployment.
- Teradata Enterprise Vector Store for integrating structured and unstructured data with generative AI models.
- Algorithm execution that connects to GPUs through Teradata AI Microservices with Nvidia.
- Data pipelines featuring data ingestion and integration capabilities with support for a variety of data types and formats.
Perhaps AI Factory's biggest benefit is the control it gives Teradata users over their AI landscape, according to Ni.
"AI Factory gives analytics leaders the levers they need to scale AI without sacrificing compliance, cost certainty, or model integrity," he said.
Some of Teradata's competitors provide similar capabilities for on-premises users, but not in the unified manner of AI Factory,, while others such as Databricks and Snowflake are almost solely focused on the cloud, Ni added.
"AI Factory is one of the first plug-and-play, governance-grade AI stacks designed for on prem, turning GPUs and data into decision engines without the cloud tax," he said. Menninger noted that Cloudera comes closest to providing an on-premises platform resembling AI Factory, although Cloudera uses third-party integrations to provide vector search and storage.
Regarding Teradata's overall offering, Teradata's technology compares well to that of its peers, but when competing for customers, the vendor suffers from a lack of brand recognition, he added.
"Their challenges have been more about marketing than they have been about technical deficiencies," Menninger said. "They compete with the hyperscalers as well as the darlings of the industry such as Databricks and Snowflake. It's not an easy task."
Looking ahead
Providing a foundation for agentic AI is a focal point for Teradata as it plans product development, according to Landry.
"We want to enable a broader set of users to get better value and more insight out of their data estate," Landry said. "That's been our mission forever, and we think this moment around agentic systems is an incredibly powerful one."
Reaching a broader set of users would be wise for Teradata, according to Menninger. But not simply by making it easier get value from data.
"The thing that would help Teradata and, indirectly, its customers, is to make Teradata cool again," Menninger said, noting that the vendor has kept up with recent trends by adding support for open table formats, open source tools, multiple processing engines and building out their ecosystem.
Ni, meanwhile, suggested that one area Teradata would be wise to focus on is AI governance as it attempts to remain among the more innovative vendors. Another would be improving the interplay between its different deployment options.
"That [would enable] more real-time intelligence, broader model integration and semantic consistency across AI and BI," Ni said.
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.
