AI will solve humanity’s toughest problems, not replace us: Larry Ellison

Oracle Chairman and CTO Larry Ellison has said that artificial intelligence (AI) would fundamentally reshape how humanity solves complex problems—from healthcare and infrastructure to agriculture and climate. Delivering the keynote at the Oracle AI World 2025 in Las Vegas, he described AI as a network of “electronic brains” that augment people rather than replace them. He called it a force “more powerful than the Industrial Revolution.”

“The real opportunity is not just building these extraordinary electronic brains, but using them to solve humanity’s most difficult problems,” Ellison said, outlining how Oracle is positioning its cloud and enterprise software stack to enable AI reasoning securely on both public and private data.

Building the infrastructure for intelligent systems

Ellison compared the human brain’s 20-watt power consumption with the 1.2-billion-watt energy demand of modern AI clusters to underline a simple truth: creating electronic brains requires industrial-scale digital infrastructure. The shift from neurons to GPUs, he said, has transformed computing into an engineering challenge that spans energy, cooling, and networking.

To meet this demand, Oracle Cloud Infrastructure (OCI) is being expanded into one of the world’s largest AI computing networks — including a new cluster in Abilene, Texas, that will eventually host more than 450,000 NVIDIA GPUs. The buildout integrates power generation, liquid cooling, and high-speed data links at a scale Ellison likened to “national infrastructure projects.”

“We are building billion-watt power plants and connecting them directly to our data centres,” Ellison said. He added that true AI capability requires more than GPUs—it depends on the reliability of the surrounding digital infrastructure. “If you do not have the data pipelines, execution environments, and low-latency systems ready, it is like buying an F1 car and parking it on a dirt road.”

Ellison said Oracle’s approach is distinct from other hyperscalers by combining massive AI infrastructure with vertical applications for industries such as healthcare, finance, and utilities. The company, he noted, is training and hosting multiple multimodal models, including those from OpenAI and xAI, while also providing customers with access to models such as Grok, ChatGPT, Llama, and Gemini in its cloud.

From public data to private reasoning

Ellison highlighted a fundamental limitation of today’s general-purpose models: they are trained almost entirely on public internet data. “ChatGPT does not know your accounts, your customer orders, or your supplier contracts,” he said. The next breakthrough, he argued, will come from enabling AI to reason over private, enterprise-owned data without compromising security.

Oracle’s new AI Database and AI Data Platform, he explained, “vectorise” data stored in Oracle and third-party clouds so that AI models can access and reason on it through retrieval-augmented generation (RAG). This privacy-preserving approach allows customers to combine public and proprietary data while maintaining full ownership and control.

The goal, Ellison said, is to move from passive information retrieval to active reasoning. “Inference is no longer about finding the right answer—it is about deciding what to do under complex conditions,” he said, emphasising that such reasoning engines could transform entire sectors.

Healthcare as the proving ground

Ellison cited healthcare as one of the most promising testbeds for AI ecosystems. Beyond modernising Cerner’s electronic health record (EHR) software, he envisioned interconnected AI agents that automate the entire healthcare value chain—from patients and providers to insurers, regulators, and financial institutions.

He illustrated how one agent could synthesise medical records, clinical research, and reimbursement policies to recommend optimal treatments that are both effective and covered under prevailing insurance rules. Another could assess verified receivables to help hospitals raise short-term financing, improving liquidity while reducing administrative complexity.

Ellison also discussed emerging innovations such as metagenomic devices capable of sequencing pathogens and identifying early-stage cancers within minutes. Such low-cost devices, he said, could one day form a global early-warning network for pandemics. He cited the example of remote patient monitoring, where IoT-connected medical devices and AI dashboards allow clinicians to track patients at home, in ambulances, and during transfers—turning care into “a continuous digital loop.”

Stressing that biometric sign-in and payments could prevent medical fraud and identity theft, eliminate passwords, and ensure privacy, he said, “Passwords are insane.” “Biometric authentication and AI-driven credit verification could make digital identity and transactions more secure”.

Another possibility he discussed was using autonomous drones with AI-based air traffic control to transport medical samples, detect wildfires, or assist emergency services—technologies he believes will make connected health and safety systems far more resilient.

Beyond healthcare: food, climate, and sustainability

Ellison expanded his vision to the planet’s larger challenges—food security, water scarcity, and carbon management. He described how AI-assisted genetic engineering could develop wheat and corn varieties that absorb more carbon dioxide and increase yields, offering a potential solution to Africa’s growing population.

He also highlighted nitrogen-fixing crops that draw nutrients directly from the atmosphere rather than fertilisers, reducing both costs and water pollution. “In Africa, most farmers cannot afford nitrogen fertiliser. In the rest of the world, it’s damaging the environment. AI can help us eliminate both problems,” he said.

Ellison described AI-automated greenhouses that use 90% less water and enable year-round cultivation near urban centres—an idea suited for countries like India, where farmland and water are limited. These automated environments, he added, could even serve as prototypes for future Martian habitats.

He also touched on the concept of biomineralisation—using AI-designed plants to convert atmospheric CO₂ into calcium carbonate, permanently locking carbon into inert form. “We can manage CO₂ in the atmosphere the same way nature builds coral reefs,” he said, framing it as an example of how AI could be applied to planetary-scale environmental management.

Ellison concluded that AI’s true impact will come not from model-building but from operational reasoning across secure, interconnected ecosystems. For this transformation to occur, he said, industries must invest in robust digital infrastructure that powers, connects, and safeguards these “electronic brains.”

“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)