AI will redefine how humanity solves problems: Larry Ellison

What happens in Vegas does not always stay in Vegas. At Oracle AI World 2025, Chairman and CTO Larry Ellison outlined how the company is reshaping its cloud and enterprise software strategy to meet the demands of the artificial intelligence era. The keynote combined two themes—the buildout of massive AI infrastructure on Oracle Cloud Infrastructure (OCI) and the deployment of AI across complex sectors such as healthcare—to show how Oracle plans to move from model-building to real-world problem-solving.

Ellison described AI as a network of “electronic brains” designed to augment human capability rather than replace it. He said the company’s focus has shifted from helping others train large models to enabling them to apply those models securely to their own data and operations. “The real opportunity is not just building these extraordinary electronic brains, but using them to solve humanity’s most difficult problems,” he said.

At the centre of Oracle’s offering is a privacy-preserving framework that lets enterprises use AI models on their most valuable data without compromising control or security. The Oracle AI Database and AI Data Platform can “vectorise” data from Oracle and third-party cloud stores, making it usable for AI models via retrieval-augmented generation (RAG). OCI will host a range of foundation models—including Grok, ChatGPT, Llama and Gemini—so that customers can combine public and private data for analysis while maintaining confidentiality.

Ellison also highlighted Oracle’s progress in “agentic” automation—applications that are generated and managed by AI itself. Much of Oracle’s new enterprise software, he said, is now created using AI and orchestrated through stateless, secure agents built with Oracle APEX. These agents can perform multistep reasoning and execute tasks such as data integration and workflow orchestration without the manual coding required by traditional enterprise systems, he said.

Building AI Infrastructure at Unprecedented Scale

Ellison said Oracle is constructing some of the world’s largest AI data centres to meet soaring demand for compute power. The OCI cluster in Abilene, Texas, now under development, will eventually include more than 450,000 NVIDIA GPUs and provide the scale needed to train multimodal models for customers worldwide. Oracle is also working with partners, including OpenAI and xAI, to train and deploy their models in its cloud.

He said Oracle’s infrastructure expansion goes far beyond data centres, involving new energy-generation systems, cooling technologies and high-speed networking to support AI workloads. “We are building billion-watt power plants, connecting them directly to our data centres,” Ellison said, noting that the scale of these projects is comparable to national infrastructure.

Oracle positions itself as distinct from other hyperscalers by combining AI infrastructure with industry-grade applications. While most large cloud providers focus mainly on generic compute and storage, Oracle’s strategy integrates large-scale AI compute with domain-specific platforms for healthcare, finance and utilities.

Ellison said the company is involved in training multiple multimodal models and expects OCI to be a key platform for enterprise-scale AI reasoning. The infrastructure is designed not only to train models but also to host AI-driven services and applications that require real-time, high-security access to enterprise data.

Reimagining Healthcare through AI Ecosystems

Healthcare emerged as one of the most ambitious areas Ellison believes AI can transform. He said the potential goes far beyond modernising existing electronic health record (EHR) systems, envisioning a future where interconnected AI agents could automate the wider healthcare ecosystem. These agents, he explained, could coordinate information and decisions among patients, hospitals, insurers, regulators, pharmaceutical firms and financial institutions that support care delivery.

He gave examples of how such systems could operate. One AI agent, Ellison said, could recommend optimal treatment plans by combining clinical data, current research and reimbursement rules. Another could compile verified receivables, enabling hospitals to raise financing against expected insurance payments and maintain liquidity. These examples, he said, illustrate how AI can manage not just information but “the entire chain of decisions and interactions that define modern healthcare.”

Ellison also cited several emerging health technologies that he believes will transform care delivery. He spoke about biometric sign-in and payment systems to prevent identity theft and fraud, and about remote monitoring tools that would allow clinicians to track patients at home or during emergency transport. He also discussed work underway on low-cost metagenomic sequencing devices to detect pathogens and early-stage cancers faster—tools that, he said, could form a global early-warning system for future pandemics.

In another example, Ellison described how autonomous drones equipped with AI-based air-traffic control systems are being tested to transport medical samples, detect wildfires and support emergency response. He said such systems demonstrate how AI can connect “intelligent devices, data and decision-making into a single network of care.”

Beyond healthcare, Ellison said AI could also help address global challenges in agriculture and climate. He cited examples of AI-designed wheat and corn that increase yields while absorbing atmospheric carbon, and nitrogen-fixing crops that eliminate the need for fertiliser runoff—technologies he said could be transformative for Africa’s growing population. He also described automated, water-efficient greenhouses capable of reducing freshwater use by up to 90%, a model suited to countries such as India, where food production must expand despite limited arable land.

Ellison concluded that AI is entering an operational phase, where reasoning and action across secure, connected ecosystems will drive its true impact. For Oracle, the convergence of massive AI infrastructure, private-data reasoning and agent-based automation defines this new phase. “AI will make us much better scientists, engineers, and doctors,” he said. “It is not replacing us—it is extending what we can achieve.”

The author was hosted by Oracle in Las Vegas to attend Oracle AI World 2025.