LAS VEGAS—As Oracle’s CloudWorld conference returned to Las Vegas this week, the question on many lips was: Since Oracle appeared to be a bit late to the generative AI party, will its aggressive focus allow it to catch up?
It appears to be something of a theme for Oracle, which was famously late to the cloud (see below). Though, to be fair, the software and database giant is responding much more quickly and very aggressively to this scorching hot trend.
Oracle is definitely going whole hog to become a serious generative AI player. The signs were everywhere at CloudWorld.
In his CloudWorld keynote (pictured), Oracle Executive Chairman and CTO Larry Ellison said generative AI is one of those watershed moments in tech. “Generative AI showed up about a year ago,” he said. “It’s changing everything. It’s certainly changing everything about Oracle.”
The brash billionaire has always been known for a bit of hyperbole, and his keynote didn’t disappoint. The oracle of Oracle said his company is “the leading provider of generative AI infrastructure” and that Oracle’s Gen2 Cloud is the “fastest and most cost-effective” platform for running generative AI workloads.
Furthermore, the company has laid out a three-tiered enterprise approach to generative AI (see below), while at the same time making some big investments. Oracle is working with several AI companies, including Mosaic ML, Adept AI, and Cohere, to help it develop and deploy generative AI applications, and it is also spending billions on Nvidia and AMD chips. Ellison said that these AI companies are using Oracle’s Gen2 Cloud to train AI models that can generate realistic images, text, and other content.
And even though it can be argued that Oracle was not the first competitor out of the gate with a generative AI plan, it has long incorporated other forms of AI in its product lineup. With an accelerated focus now on generative AI, the company could indeed have a strong position as the market grows. According to a recent post from Steve Miranda, Oracle’s executive vice president of applications development:
“Oracle is the only technology vendor that offers both a full suite of cloud applications and a next-generation cloud infrastructure specifically designed to run them. This unique combination powers new capabilities as well as traditional and generative AI features that are embedded within Oracle Fusion Cloud Applications and creates a continuous feedback loop that drives the fastest innovation cycle in the industry.”
Oracle’s Famous Cloud Pivot
More on Oracle’s generative AI plans in a bit. But first, some history.
In the generative-AI arena, one can’t help but see a parallel to the early days of cloud computing. In those days, Ellison famously dismissed the cloud. In 2008, he said that cloud computing was “gibberish” and that “it’s going to go away.” Oracle would never build a cloud platform, he said. His change of heart, however, is a Silicon Valley legend. By 2012, he said that cloud computing was “here to stay” and that Oracle would be a major player in the cloud market.
Because of that delay, Oracle is far behind the cloud leaders Amazon Web Services, Microsoft Azure, and others. Although, to be fair, some customers we’ve interviewed say that Oracle Cloud Infrastructure (OCI) has some advantages in performance and scalability. Also, Oracle, after its most recent earnings call, says it’s the “fastest-growing” cloud provider.
Fast forward to 2020. Just like cloud computing, Ellison dissed the term “artificial intelligence.” Instead, he called it “adaptive intelligence.” He argued that the term “artificial intelligence” was misleading. “Adaptive intelligence” was a more accurate description of the capabilities of AI systems, he argued.
At the time, when this analyst heard Ellison utter “adaptive intelligence,” it seemed insignificant. Some AI experts praised him for trying to clarify the meaning of AI, while others criticized him for creating a new term that was just as confusing as the old one.
But then, OpenAI’s ChatGPT changed the landscape in late 2022. And once again, it appears that Oracle is playing catchup, albeit aggressively, after its rivals have already poured billions into their gen AI products and services.
Three Tiers
Oracle provided more details on its three-tiered approach at CloudWorld, its annual confab for users, customers, and partners. The approach “resembles a three-layer cake,” Oracle executives say. Each layer targets different enterprise customers depending on their needs.
The first tier consists of its OCI Supercluster service. This is targeted at companies, such as Cohere or Hugging Face, that are developing large language models (LLMs) to support their customers. Earlier this year, Oracle announced an investment in Cohere, a startup founded by ex-Google engineers.
OCI’s Supercluster service includes OCI Compute Bare Metal and a choice of high-performance storage options. Oracle says the AI supercomputing service can support thousands of bare metal servers with tens of thousands of Nvidia A100 GPUs for processing massively parallel applications. The service also comes with Nvidia’s foundation models, such as Nvidia Picasso and BioNeMo, along with AI training and governance frameworks.
Oracle’s second tier targets companies—including its own OCI customers—that seek to develop generative AI capabilities based on their own data for their own consumption.
Generative AI Embedded in Fusion, NetSuite Apps
The database giant is also adding generative AI capabilities to its Fusion Cloud and NetSuite applications by embedding Cohere’s foundational models into its software-as-a-service (SaaS) offerings. Metadata from these apps will be combined with foundational models to offer gen AI assistants inside these apps to boost worker productivity.
In its Fusion Cloud HCM solution, these capabilities promise to help boost human resources productivity. Capabilities include “Assisted Authoring,” which can help an HR department generate content, such as job descriptions or job postings. Another feature, “Suggestions,” can help with tasks such as automated recommendations for survey questions. Another gen AI capability, “Summarization,” pulls data together, including feedback on an employee from throughout a given year. It then summarizes that data for a performance review.
Using its Own APEX Code-Generating Platform
Oracle is also dining at its own gen AI table. In his CloudWorld keynote, Ellison said that none of Oracle’s new applications—he stressed the words “new applications”—will be built in Java. Instead, Oracle will use its own code-generating APEX platform for that. APEX, or Application Express, is a low-code application development platform that enables users to build and deploy scalable, secure web and mobile apps. It is a fully supported, no-cost feature of Oracle Database, and runs as a part of it.
One area where this approach is being used is Oracle’s Cerner acquisition, the biggest acquisition in company history. Oracle is in the process of rewriting the Oracle Cerner electronic health record (EHR) system, adding capabilities, and giving it a modern, cloud-based foundation.
Oracle is “rewriting Cerner a piece at a time,” but it isn’t rewriting the code in Java, Ellison said. Instead, it is using APEX. He said that this is a “two-phase process” with the first phase being to “harden the existing system” and move customers to the cloud. The second phase is to replace features of the older Cerner system with new features written in APEX. It is an ambitious undertaking.
Big Strides in Burgeoning Market
Oracle is definitely taking major strides and laying out some serious cash as it pursues a foothold in the burgeoning generative AI market. It’s a major focus, Oracle CEO Safra Catz said:
“We are investing billions of dollars in Nvidia chips this year. We are also spending three times that on CPUs from Ampere and AMD. We still spend more money on conventional compute, but we are investing heavily in AI. We believe that AI is the future of computing, and we want to be the leading provider of AI infrastructure.”
Oracle is certainly looking to avoid the same approach it took with the cloud, where it fell behind early, only to find it challenging to grab market share from the leaders.
But will it succeed? The amount of effort and cost Oracle is pumping into generative AI is either impressive … or possibly worrying, depending on one’s perspective. The proof will indeed be in the pudding.
The fact that Oracle offers both a full suite of cloud applications and a next-generation cloud infrastructure could be a big advantage. However, what it is offering in the conception and planning stages may be similar to offerings from rival public cloud companies. It remains to be seen how these efforts mature and if Oracle can be a true market leader in the generative AI wars.
By Tom Dunlap, Research Director, Avasant