Teradata, the provider of the most complete cloud analytics and data platform for AI, is unveiling Teradata AI Factory, a purpose-built, integrated solution for delivering Teradata’s cloud-based AI/ML capabilities to on-prem deployments. With a combination of software, hardware, and pre-packaged Teradata and modern third-party tools, Teradata AI Factory allows organizations to benefit from advanced AI while maintaining security in their on-prem environments.
If there is one constant, it’s that the landscape of AI is continuously evolving, bringing to light exciting new innovations—as well as new concerns. Today, many organizations seek the benefits of AI, yet are inundated with high-risk data, strict regulatory policies, and tight budgets.
Offering additional context for this launch, Dr. Meeta Vouk, vice president of product management and AI and analytics at Teradata, explained how, “We have a lot of customers that have [stringent] data governance, data regulation requirements, and extremely sensitive data. We… [call ourselves the] custodians of the world's mission critical data…Our customers on-prem were also asking for…[advanced] AI capabilities for their extremely regulated data.”
“We've seen tremendous innovation with AI happening in the cloud; … [Our customers] wanted that innovation to be on-prem as well,” Dr. Vouk continued.
Teradata AI Factory—built with the NVIDIA Enterprise AI Factory validated design—unlocks the possibilities of predictive, generative, and agentic AI for on-prem deployments by unifying a variety of critical components for execution. Centralizing data pipelines, algorithm execution, and software infrastructure into a single, scalable system, Teradata AI Factory simplifies private, trusted AIOps with the security and cost benefits of on-prem infrastructure.
Furthermore, when leveraged in combination with Teradata AI Microservices with NVIDIA NeMo and customer-provided GPUs, AI development—including that of native retrieval-augmented generation (RAG) pipelines—is greatly accelerated.
When using Teradata AI Factory, sensitive data stays within the boundaries of an organization’s infrastructure, ultimately reducing risks and compliance issues that might otherwise be associated with public AI platforms. The platform adheres to several compliance standards, including GDPR and HIPAA, making it ideal for industries such as healthcare, finance, and government.
On top of its privacy boons, Teradata AI Factory persists Teradata’s innovation in delivering the best performance for the price—made possible through its localized architecture that reduces data movement, lowers latency, and improves efficiency. Customers can run AI models on CPUs or accelerate them using customer-provided GPU infrastructure, eliminating unpredictable cloud pricing and maximizing existing hardware investments. Additionally, customers can bring any model to Teradata AI Factory, ensuring flexibility and choice.
Finally, Teradata AI Factory emphasizes simplicity and ease-of-use, where the seamless integration of hardware and software removes the need for complex custom setups. This ultimately expedites AI development and deployment while reducing operational complexity, according to Teradata.
"Market dynamics are increasing buyer interest in on-premises solutions,” said Teradata’s chief product officer, Sumeet Arora. “Teradata remains the clear leader in this environment, with proven foundations in what makes AI meaningful and trustworthy: top-notch speed (performance), predictable cost (resource efficiency), and integration with the golden data record (which may already live on Teradata). Teradata AI Factory builds on these strengths in a single solution for organizations using on-prem infrastructure to gain control, meet sovereignty needs, and accelerate AI ROI.”
Additionally, the Teradata AI Factory includes:
- AI platform for rapid innovation on-prem, including the Teradata Enterprise Vector Store for integrating structured and unstructured data in GenAI applications
- Robust software infrastructure, including a self-service workspace with access to libraries and functions; built-in support for model lifecycle management, compliance, and one-click LLM deployment; as well as tools such as JupyterHub, ModelOps, Airflow, Gitea, and Devpi to ease deployment and developer workflows
- Predictive and generative algorithm execution at scale with high-performance platform analytics and AI workloads by connecting to customer GPUs via Teradata AI Microservices with NVIDIA, further enabling fast model development and native RAG processing
- Efficient data pipelines with modern data ingestion tools and Teradata’s internal solutions, including QueryGrid, as well as support for Open Table Formats (OTFs), object stores, and NVIDIA tools for complex formats, such as PDFs
“We're not just building AI for AI’s sake,” said Dr. Vouk. “[We’re] dealing with the world's mission critical workloads…[and] we are building AI with that in mind. We have the world's most sensitive data living on Teradata. What does it mean to bring AI capabilities to that?”
“We're really delivering trusted AI because we are thinking across that spectrum,” Dr. Vouk concluded.
To learn more about Teradata AI Factory, please visit https://www.teradata.com/.