An integrated data architecture that considers both security and scalability is needed.
Tirathankar Lahiri, senior vice president of Oracle's artificial intelligence (AI) engine institutional sector, stated accordingly on the 3rd at the "Oracle AI Summit 2026" held at the Grand Intercontinental Seoul Parnas in Gangnam-gu, Seoul.
Vice President Lahiri set Oracle's goal as a leap to a corporate AI platform. Lahiri said, "We focused on embedding AI at the core of the data platform, as it has become a strategic imperative in corporate data management," adding, "We are working to let customers choose where to place AI models and vector embeddings, and to support integrated AI deployment across public cloud, private cloud, and even air-gap (a physically isolated environment) options."
Vice President Lahiri put forward a "converged data architecture" as a strategy to reduce data operation expense. The intent is to help corporations run data and AI within a single architecture, rather than attaching them separately, to lower the total cost of ownership (TCO) of data management.
This "converged" message became feasible through Oracle's AI database "26 ai," launched in Oct. last year. 26 ai is a database engine with AI built in by default. Oracle has emphasized that, through vector search and retrieval-augmented generation (RAG), it can quickly find internal corporate data and combine the results with a large language model (LLM) to produce more accurate answers.
Vice President Lahiri explained, "AI vectors arrange complex meanings as sequences of numbers, but they can be generated based on the model a user selects, and we added a 'vector index' to speed up search." Lahiri added, "Through RAG, you can find meaningful documents and enable web applications to use them based on the data," and "when combined with an LLM, you can use databases APIs, and by using SQL (a language used to query and modify databases), you can quickly search for the most relevant documents."
Vice President Lahiri stressed that AI functionality is possible even when data types differ. Lahiri said, "Oracle's goal is to help make very complex AI logic easy with clear SQL," adding, "Think of a short SQL statement as equivalent to hundreds of lines of Python code." Lahiri explained, "Not only structured data, but other types of data can also be combined and expressed with SQL statements."
Oracle is also expanding collaboration with global AI models. Vice President Lahiri said, "Through collaboration with global leading AI model corporations such as xAI, Google, Meta, Cohere, and OpenAI, we are targeting corporate customers with powerful AI models and Oracle's specialized AI agent strategy." Oracle has adopted Oracle Cloud Infrastructure (OCI) to accelerate AI model development and deployment and has built an environment that can strengthen competitiveness in global markets. Oracle has 430,000 customers using OCI.
Oracle Korea is preparing a business partnership with Amazon Web Services (AWS) in the first half of this year. Kim Seong-ha, head of Oracle Korea, said, "Exadata (a databases-dedicated platform that Oracle optimized as a combined set of hardware and software) equipment has been installed in the domestic AWS data center. We are actually doing the prep work now," adding, "A domestic digital financial company is conducting a proof of concept (PoC) for service adoption by using Oracle data centers in a U.S. region."