LAS VEGAS – Oracle chairman and chief technology officer Larry Ellison has never been one to think small. At Oracle AI World 2025, the 81-year-old billionaire painted a sweeping, often audacious picture of artificial intelligence as “the largest, fastest-growing business in human history”, comparing its impact to the Industrial Revolution and the birth of the internet, and putting Oracle at the centre of it all.
“People say AI changes everything,” Ellison told a packed ballroom. “That is a big statement, everything, but I think it is pretty close.”
Ellison’s 90-minute keynote fused technical explanation with Silicon Valley showmanship. He argued that AI training has already eclipsed every previous industrial investment cycle, including the building of railways and the electrification of cities. The proof, he said, lies in the physical scale of Oracle’s new data infrastructure.
In Abilene, Texas, Oracle is constructing what Ellison called “the world’s largest AI cluster”, a sprawling complex spanning more than 1,000 acres and expected to house roughly half a million NVIDIA GPUs. When fully powered, it will draw 1.2 billion watts of energy, enough to run a city of one million homes.
“It is a long way from writing code in my bedroom in college,” Ellison said. The site, co-developed with OpenAI, is being built at a pace more typical of oil refineries than cloud data centres. More than 3,500 workers are on site daily. Power is sourced from both the local grid and on-site natural-gas turbines.
To Ellison, such investments represent a historic inflection point. “We are not building 20-watt meat computers, we are building 1.2-billion-watt electronic brains,” he said. “A whole new world is dawning.”
Private data, public models
Beyond the spectacle of power plants and GPUs, Ellison outlined how Oracle plans to make itself indispensable in the AI ecosystem, not by building its own consumer models but by enabling others to use enterprise data securely with them.
Most corporate data, he reminded the audience, already sits in Oracle databases. The company has reengineered that software to work with large language models through a technology known as retrieval-augmented generation (RAG). The system allows AI models such as OpenAI’s ChatGPT, Anthropic’s Claude or Elon Musk’s Grok to reason over private corporate data while keeping it siloed.
“The biggest opportunity is not training models,” Ellison said. “It is using them to solve humanity’s most difficult problems.” He described Oracle’s “AI Database” as a bridge between public internet models and private enterprise intelligence. “The model does not have to be retrained; it just needs to access the right data securely.”
Oracle, he said, already uses AI agents internally to predict which customers are most likely to purchase new products and to automatically generate personalised sales messages. “A lot of our code is written by our AI models now,” he said.
Ellison, whose company acquired health IT giant Cerner in 2022, reserved his most detailed examples for healthcare, a sector he called “the ultimate test of AI’s promise”. Oracle is rebuilding Cerner’s software with AI-generated code and plans to modernise not just hospitals and clinics but the entire ecosystem connecting patients, providers, insurers, regulators and even banks.
Also Read | Oracle CEO says ‘AI changes everything’ as company bets on ‘trusted AI’
He described a scenario in which an AI agent assists a doctor in selecting the best treatment based on medical records, reimbursement rules and patient affordability, even checking clinical trial databases in real time. “The goal is the best possible care that is fully reimbursable,” he said. “AI can help ensure no patient is denied care because of paperwork or outdated rules.”
Another Oracle pilot uses AI to assess hospitals’ outstanding insurance claims, giving banks confidence to lend against receivables, potentially easing liquidity problems that cause care delays.
Ambitions that stretch beyond software
The keynote extended well beyond cloud computing. Ellison spoke enthusiastically about prototypes ranging from AI-assisted imaging devices and metagenomic diagnostic tools to autonomous ambulances transmitting live patient data to hospitals.
He also described ventures in robotics and agriculture. Oracle and its partners, he said, are building robotic greenhouses capable of producing food with 90% less water and potentially doubling yields through AI-designed crops that absorb more carbon dioxide. The technology, he suggested, could one day support colonisation efforts on Mars. “I told Elon it doubles as a Martian habitat,” he joked.
He also cited “Wild Bio”, a company backed by Oxford University, which is experimenting with genetically engineered wheat to increase CO₂ capture, positioning AI as a tool for climate remediation as much as computation.
In logistics, Ellison highlighted Oracle’s drone projects that deliver medical samples, detect wildfires and track vehicles to replace police pursuits. “We built an air-traffic control system for drones,” he said. “It keeps people safe and preserves privacy.”
While Ellison dismissed fears that AI would replace humans, “it will make us better scientists, engineers, teachers and chefs”, his broader message underscored the sheer concentration of capital and energy now driving AI development. With power-hungry clusters and data pipelines running through Oracle’s cloud, the company hopes to become the infrastructure layer for the world’s AI ambitions.
Rivals like Microsoft, Amazon and Google are also investing heavily in AI compute but Ellison sought to distinguish Oracle by claiming it uniquely combines infrastructure with enterprise applications. “We build the hardware and the software to automate entire industries,” he said. “Others just provide the infrastructure.”
For Oracle, a company once defined by its database dominance, the shift to AI infrastructure and healthcare modernisation marks both a survival strategy and a return to form. It is betting that as AI models become commodities, data integrity and compute efficiency will matter more than branding.
For many in attendance, from people eager to enter the AI workforce to CIOs juggling cloud costs, Ellison’s message landed clearly. The next industrial revolution, he insisted, will not be written in code alone but powered by gigawatts, terabytes and the stubborn human belief that machines can help build a better world.
