Our team was on the ground in Las Vegas for the first rebranded edition of Oracle AI World (formerly CloudWorld).
The headline: Oracle left the usual hype behind and focused on what CIOs and IT leaders are really trying to solve right now. As Larry Ellison put it,
“Bring the model you want, train it with public and private data, and keep it private and secure.”
That line perfectly captured Oracle’s focus this year. The company has been working hard to tackle what I like to call the winning triangle: how to use AI with private (sensitive, high-value) data, how to control who sees what, how to scale globally within regulations, and how to power it all with the right infrastructure and energy footprint.
In other words, less speculation, more execution.
As Steve Miranda, Executive Vice President of Oracle Applications Development, said during his keynote:
“I’m not here to talk about AI. I’m here to show you AI.”
Oracle’s approach felt like the natural post-hype phase of AI – treating it as another essential layer of the enterprise stack, like cloud, ERP, or CRM. A technology whose potential, limits, and risks are now well understood, and that’s being engineered into the everyday operations of business around the world.
The end of the hype cycle: how Oracle is making AI part of everyday business
This year’s edition stood out for its pragmatic tone. Oracle left behind the inflated promises that often surround AI and focused on what’s actually working. Throughout the sessions, the message was clear: AI is no longer a future add-on or a “bolt-on” feature. It’s now part of how enterprises operate every day.
Larry Ellison illustrated this perfectly when he explained that a growing portion of Oracle’s own codebase is now written by AI. Developers define the intent, and the system generates the code automatically. That approach allowed Oracle to rebuild Cerner’s 25-year-old healthcare platform in just three years.
In many ways, Oracle captured where the industry stands today in Gartner’s Hype Cycle: the Trough of Disillusionment. The term might sound negative, but it actually marks a stage of maturity. It’s the point where the initial excitement gives way to understanding. AI is no longer the shiny new thing; its possibilities, limits, and risks are better understood, and it’s becoming another essential layer of enterprise infrastructure, just like cloud, CRM, or ERP did once.
Day 1
1) Mike Sicilia Keynote – Oracle AI: Powering Your Business
Mike Sicilia, Executive Vice President of Oracle Global Industries, opened Day 1 with a grounded message about AI’s real-world impact.
“AI is an enabler,” he said. “It’s helping people do their work better, not replacing them.”
He cited Oracle’s latest research showing that 80% of employees in organizations using AI reported a measurable improvement in the quality of their work. That, Sicilia emphasized, is the metric that really matters.
The session was centered on how AI is already transforming core operations across industries — from healthcare and energy to mobility and life sciences. Sicilia invited four Oracle customers to the stage, each representing a different sector, to share how they’re using AI to solve concrete business challenges:
Exelon – Predicting and Preventing Grid Failures
Calvin Butler, CEO of Exelon, described how the energy provider uses Oracle AI to predict grid failures and maintain reliability for millions of Americans. With a network that spans six states and serves more than 10 million customers, Exelon’s AI models analyze weather patterns, equipment conditions, and historical data to anticipate outages and dispatch maintenance crews before problems occur. The result: greater efficiency, lower costs, and safer infrastructure.
Avis Budget Group – Augmenting Decision-Making
Ravi Simhambhatla, CIO of Avis Budget Group, explained how Oracle Database 23AI lets employees query data in plain language and automate procurement decisions. Instead of coding complex queries, team members can simply ask questions like “Which vehicle segments show the highest margin growth?” and receive accurate answers instantly. Simhambhatla emphasized that the goal is to empower every worker, not replace them: “AI doesn’t stand for Artificial Intelligence — it stands for Augmenting Individuals,” said Ravi Simhambhatla.
Marriott International – Integrating Hospitality Systems
Ty Breland, EVP and Chief Human Resources Officer of Marriott International, outlined how the company is using AI and Oracle Fusion Applications to connect disparate systems across its global portfolio. By consolidating dozens of property-management, HR, and finance platforms into one AI-powered interface, Marriott enables frontline employees to spend less time on manual data entry and more time creating memorable guest experiences.
Biofy (Brazil) – Accelerating Healthcare Diagnostics
Finally, Paulo Perez, CEO of Biofy, a biotech company based in Brazil, shared how Oracle’s vector database is helping detect bacterial resistance and recommend targeted treatments in a fraction of the time. By analyzing genetic sequences and clinical data through AI-powered search, Biofy cut diagnostic time from five days to just four hours — a life-changing difference for patients and providers alike.
2) Larry Ellison Keynote – Oracle Vision and Strategy
Larry Ellison, Oracle’s Chairman and CTO, opened Oracle AI World with an unfiltered look at where AI really stands today — beyond the buzz, deep in the engineering. “AI changes everything,” he said early on, calling it “one of the greatest inventions in human history, bigger than the railroads or even the Industrial Revolution.” But what defined his talk wasn’t grand predictions — it was the hard work behind making those words true.
The two biggest opportunities with AI
Ellison separated the AI boom into two clear phases:
- Building and training massive models, the stage where the tech giants and startups alike have invested billions to turn raw internet data into intelligence.
- Using those models for reasoning — making them act, decide, and solve real-world problems.
That second phase, he said, is where Oracle sees the biggest opportunity. “The real change won’t come from building the models, but from using these remarkable electronic brains to solve humanity’s most difficult and enduring problems.”
He explained that while every major model (ChatGPT, Claude, Gemini, Grok) is trained on public data, the next frontier lies in reasoning securely with private data, the kind that lives inside enterprises and carries the most value. That’s exactly where Oracle’s strategy comes in.
The private data problem and Oracle’s solution
Ellison outlined a major paradox: organizations want AI to think over their sensitive data but can’t risk exposing it. “People want to have their cake and eat it too,” he said. “They want models to reason on their private data, but still keep it private.”
Oracle’s answer is the AI Data Platform, powered by the new AI Database, built to enable retrieval-augmented generation (RAG). Instead of training models on private data, Oracle makes that data available for reasoning through vectorization, converting it into embeddings models can understand while keeping access tightly controlled.
In practice, it means customers can connect any multimodal model to both public and private data, all within a secure Oracle environment. “The model can reason across your data, not look at it,” Ellison emphasized.
AI that writes Oracle’s code
One of the most striking moments of the keynote came when Ellison revealed that a growing portion of Oracle’s own codebase is now written by AI. “We don’t write the procedure anymore,” he said. “We declare our intent, and the model writes the step-by-step process.”
This approach, built into Oracle APEX with its declarative AI generation language, has already proven itself because it enabled their team to modernize Cerner’s 25-year-old healthcare system in just three years. The AI-generated applications are not only faster to build but also secure, stateless, and designed to scale for millions of users from day one.
For Ellison, this isn’t AI as a “bolt-on”; it’s AI built into the DNA of how Oracle engineers work.
The world’s largest AI data center
Ellison also unveiled Oracle’s massive AI infrastructure project: a 1.2-billion-watt data center in Abilene, Texas, built in partnership with OpenAI. When complete, it will house over 450,000 NVIDIA GPUs, making it the largest AI cluster ever constructed.
He joked about the project’s scale: “That’s a long way from writing code in my bedroom in college… what happened? No idea.”
AI beyond software: healthcare, security, and everyday life
Ellison closed with a glimpse into the future applications Oracle is already pursuing:
- Healthcare: AI-powered medical imaging, real-time diagnostics, and genomic testing that can detect cancer in hours instead of weeks.
- Security and Identity: Eliminating passwords through biometrics to prevent identity theft.
- Connected Devices: Smart ambulances and medical devices that feed data directly into secure AI databases for real-time monitoring.
Day 2
1) Clay Magouyrk Keynote – Building the Cloud for You
Day 2 picked up right where Larry Ellison left off: less talk about what AI could be, more focus on how it actually works.
Clay Magouryk, Executive Vice President of Oracle Cloud Infrastructure (OCI), took the stage to explain what’s behind the scenes: the networks, chips, and architecture that make everything Ellison described possible. It was the kind of session that pulled back the curtain on the plumbing of AI; the part most people don’t see, but without which none of it would run.
TikTok: creativity at planetary scale
The session opened with a fireside-style conversation between Magouryk and Fong-Fei Chan, Head of Infrastructure Engineering at ByteDance (TikTok). It was one of the most tangible examples of scale and collaboration in the entire event.
TikTok runs on millions of servers and moves hundreds of terabytes per second in network traffic — all built in partnership with Oracle Cloud Infrastructure (OCI). Together, they’ve pushed the boundaries of connectivity, even pioneering the industry’s first 100G and 400G FastConnect.
Chan described TikTok as “the congress for people to create, the window to discover, and the bridge to connect,” emphasizing how infrastructure directly enables creativity. The growth figures were staggering: a 60% increase in active users since 2021, now surpassing 1 billion worldwide.
Behind those numbers lies a shared obsession with reliability. TikTok and Oracle run joint stability goals at the leadership level, ensuring both teams treat uptime as a shared responsibility, not a contractual SLA. “We write our operational procedures together,” Chan noted.
Introducing Acceleron: the engine beneath OCI
After showcasing TikTok’s infrastructure story, Magouryk announced one of the event’s most important engineering updates: Acceleron.
Acceleron is OCI’s next-generation input/output architecture, combining software, hardware, and security enhancements to deliver faster throughput, lower costs, and tighter protection for every workload. It reflects years of work on network fabrics, smart NICs, and disintermediation — the removal of “middle boxes” that slow down data flow.
In simple terms, Oracle is rethinking how data moves through the cloud. Instead of layering complexity, Acceleron strips away friction and enables a multi-plane network that’s faster, safer, and easier to manage. It’s the invisible backbone behind everything from high-frequency trading to AI training clusters.
OpenAI and the race for compute
The second conversation of the session was with Peter Hoeschele, VP of Infrastructure and Industrial Compute at OpenAI.
Hoeschele described the partnership with Oracle as a turning point. “When we needed to scale fast, Oracle didn’t just provide hardware — they co-engineered with us,” he said. “They understood what we were trying to achieve and moved at the same pace we did.”
Together, OpenAI and Oracle are building some of the world’s most powerful clusters, capable of training and serving next-generation models in real time. Hoeschele called this the industrialization of compute, a shift from isolated data centers to interconnected, intelligent networks that can dynamically adjust power and resources as AI models evolve.
“We’re past the point of separating training and inference,” he noted. “Models are now continuous systems — always learning, always reasoning — and that changes everything about how we design infrastructure.”
From engineering to enablement
Magouryk closed the session by tying it all together.
He introduced Oracle’s new AI Data Platform, combining the AI Database, GenAI Services, and Agent Platform into one ecosystem. This platform allows enterprises to reason securely across their private data without exposing it — connecting the dots between Ellison’s Day 1 message about privacy and Magouryk’s Day 2 engineering deep dive.
Magouryk summed it up best:
“AI is only as good as the data it has access to. Our job is to make that access secure, fast, and universal.”
2) Steve Miranda Keynote – Accelerate Success with AI in Oracle Applications
Steve Miranda, Executive Vice President of Applications Development at Oracle, closed the event with a simple statement that summed up the company’s attitude toward AI:
“I’m not here to talk about AI. I’m here to show you AI.”
And he did exactly that. His session was a series of live demonstrations showing how AI is already embedded across Oracle Fusion Applications and turning everyday workflows into intelligent, automated processes.
In finance, he showed how AI can reconcile transactions, flag anomalies, and generate reports without manual input. In HR, he presented how it can answer employee questions and fill out forms using historical data. In supply chain, it predicted delays and suggested alternative suppliers before disruptions occurred.
Behind these examples lies the same architecture Clay Magouryk introduced earlier: the AI Database and the Agent Platform. These tools use retrieval-augmented generation (RAG) to combine the intelligence of public models with the security of private enterprise data. That’s what makes Oracle’s approach different, making AI work directly within the applications companies already use.
Miranda stressed that this isn’t about replacing people, but augmenting decision-making and removing friction. “You shouldn’t have to move your data into an experimental sandbox just to get value from AI,” he said. “It should work where your data already lives.”
The session also highlighted how developers can create their own AI agents within Oracle’s ecosystem. Using a declarative, low-code process, teams define intent and permissions, and the platform takes care of the orchestration and security.
Conclusion
As Oracle partners, it’s always valuable to hear directly from the people shaping the company’s roadmap. Events like Oracle AI World give us a clear view of where the technology is heading, and how we can help clients move faster with the right talent.
The takeaway from this year’s edition is simple: AI is already embedded on how businesses operate. The conversation is no longer about potential use cases, but about how to execute safely, efficiently, and at global scale.
Oracle is tackling the toughest questions companies face when adopting AI:
- how to work with private data without putting it at risk,
- how to control who can access what,
- and how to meet regional compliance requirements without slowing down innovation.
Oracle has built a foundation that makes these challenges manageable.
And that’s where we come in. At Inclusion Cloud, we help enterprises deploy Oracle solutions with certified professionals who understand both the technology and the business goals behind it. If you’re planning your next step with AI, we can help you make it real.
