Oracle AI World 2025 was held in Las Vegas from October 13 to 16, 2025. The annual Oracle AI event drew tens of thousands of developers, business leaders and industry experts. Under the theme "AI as the Operating System," the conference focused on how AI is reshaping enterprise digital transformation. Oracle founder and CTO Larry Ellison emphasized in his opening keynote that AI is no longer an auxiliary tool but a core driver of enterprise infrastructure. The event highlighted the latest AI innovations on Oracle Cloud Infrastructure (OCI), stressing open source, integration and security. According to Oracle's official blog and TechCrunch reports, the event marks Oracle's transition from a database giant to a full‑stack AI provider and is expected to drive AI‑related revenue growth of more than 30% in fiscal 2026.
Key Takeaway 1: Comprehensive upgrade of OCI AI agents
The centerpiece of the conference was the launch of OCI's AI agents ecosystem. Ellison announced that OCI will integrate multimodal AI agents to automate complex business processes such as supply‑chain optimization, customer service and financial forecasting. These agents are built on Oracle's Vector Search and Generative AI services and can handle both structured and unstructured data. ZDNet reported that the new tools include "Autonomous AI Agents" capable of self‑learning from enterprise data lakes and enabling zero‑code deployment. For example, within ERP systems, an AI agent can analyze sales data in real time to predict inventory shortages, improving accuracy by 25%. Oracle emphasized that these agents are compatible with open‑source models such as Llama and Mistral to avoid vendor lock‑in. An X platform user, @OracleCloud, shared a demo video showing agents handling 1,000+ queries per second in a simulated environment. This upgrade directly addresses enterprise demand: Gartner predicts that by 2027, 80% of enterprises will adopt AI agents, positioning Oracle as a leader.
Key Takeaway 2: Deep partnerships with NVIDIA and AMD to advance AI infrastructure
Oracle significantly strengthened partnerships with AI hardware vendors. Ellison revealed that collaboration with NVIDIA is expanding to the OCI Supercluster, offering up to 65,000 Blackwell GPUs to support training of trillion‑parameter models. The cost efficiency is notable: training a model similar to GPT‑4 on OCI reportedly requires only 70% of the cost on traditional clouds. At the same time, AMD's Instinct MI300X accelerators are being integrated to support hybrid cloud AI workloads. VentureBeat analyzed that these collaborations aim to challenge AWS and Azure's dominance, and OCI's AI compute capacity is now among the top five globally. The conference demonstrated an "OCI AI Infrastructure as a Service" option allowing enterprises to lease dedicated AI clusters, with entry pricing as low as $0.5/hour/GPU. Ellison emphasized sustainability: OCI data centers use liquid cooling technology, reducing carbon emissions by 40%. This takeaway highlights infrastructure democratization benefiting small and medium enterprises.
Key Takeaway 3: Industry‑specific AI solutions deployed
The conference emphasized vertical industry applications. In healthcare, Oracle Health AI introduced predictive analytics tools that use AI agents to screen for diabetes complications with 95% accuracy and are piloting with the Mayo Clinic. In financial services, OCI Fraud Detection AI integrated blockchain to detect money‑laundering in real time, reducing false positives by 50%. Retail giant Walmart shared a case in which Oracle AI optimized its supply chain, improving inventory turnover by 15%. SiliconANGLE reported that manufacturing solutions include digital‑twin AI capable of simulating factory operations and predicting equipment failures. Oracle committed that all solutions comply with GDPR and CCPA privacy regulations and support data sovereignty. X user @LarryEllison quoted in his speech: "AI is not about replacing humans but about amplifying human intelligence." This section emphasized practical AI applications rather than hype.
Key Takeaway 4: Developer tools and open ecosystem push
For developers, Oracle released an AI DevKit that includes a low‑code platform and a marketplace for pre‑trained models. Key tools such as the Oracle APEX AI Extension enable drag‑and‑drop AI app building without programming experience. The conference announced that parts of the OCI Generative AI service code will be open source and compatible with the Hugging Face model hub. InfoWorld noted that this move could attract 100,000+ developers to the Oracle community. Additionally, the Oracle Code Assist AI coding assistant was upgraded to support real‑time debugging for Java and Python, doubling developer productivity. Ellison revealed that an AI security sandbox will be launched in 2026 to prevent model hallucinations and data leaks.
Key Takeaway 5: Security, ethics and future outlook
Security was a pervasive theme throughout the conference. Oracle introduced an AI Guardrails framework to automatically detect bias and toxic outputs, aligning with the EU AI Act. Ethics panels featured guests including Anthropic CEO Dario Amodei, who emphasized responsible AI. In his closing remarks, Ellison projected that by 2030 AI will contribute 50% of Oracle's revenue and add $15 trillion to the global economy. CRN reported that Oracle's stock rose 5% during the event, reflecting market confidence.
Through this conference, Oracle AI World 2025 reinforced Oracle's AI strategy: full coverage from cloud infrastructure to application layers, emphasizing integration and practicality. The takeaways not only showcased technical advances but also responded to enterprise pain points such as cost and security, driving AI from experiments to scaled deployments. Looking ahead, Oracle may become the "operating system" for enterprise AI, but it will still face competition and regulatory challenges.