Enable ubiquitous AI workloads; open engines and formats attain parity.

By 2025, the cost performance of pluggable open engines to process open-format data will reach parity with proprietary online analytical processing solutions for many use cases. We will extend our data platforms with, and for, generative and agentic AI.

Why this matters for our clients and the world

Enterprises will use AI-driven tools such as text-to-SQL to prepare and access data products simply and securely. The growth in techniques to process and understand multi-modal data will increase the value that businesses will extract from their data.

The technologies and innovations that will make this possible

We will use AI models and agents for better natural language to discover, transform, validate, and gain insights from structured and multimodal data. We will extend both enhanced security in vector stores and ingest and RAG data pipelines with enterprise features that handle structured (in and across documents), unstructured, and multimodal data and preserve source entitlements. Open engines will leverage GPUs for query and data frame acceleration.

How these advancements will be delivered to IBM clients and partners

Text-to-SQL and multi-modal data support will be integrated as common components into a range of products. Unstructured data integration will deliver advanced extraction from documents to multiple repositories. Data governance compliance will be offered using code-driven processes instead of post-fact forms. The watsonx.data open lakehouse will advance cost-performance parity.