“A lot of other cloud vendors have been let off the hook”: Oracle leans hard on one-size-fits-all appeal of OCI for enterprises

Oracle is betting big on the scalability and reliability of its cloud offerings, as it looks to convince customers that off-the-shelf AI agents grounded in secure data can meet business needs better than the bells and whistles of its competitors.

In the opening keynote at Oracle AI World Tour London, Oracle CEO Mike Sicilia said mass adoption of generative AI has rapidly accelerated the pace of innovation and demand for new breakthroughs.

“Expectations in the age of AI haven't simply risen,” Sicilia said, “they've reset entirely and remarkable things that felt out of touch just a few years ago, a few months ago, and in some cases just a few weeks ago, are now the new normal.”

To meet this demand, he said, Oracle is looking to reengineer the cloud behind the scenes so that enterprises can adopt, deploy, and manage their data and enterprise AI tools as easily as possible.

Nothing Sicilia said on stage would have come as a great surprise to attendees. In fact, the most surprising thing to me throughout the keynote was the confidence with which Oracle is promoting its ‘stay the course’ approach to cloud and AI in 2026.

Scalable cloud for reliable AI

Oracle’s pitch is simple: scale and stability. Rather than becoming overcomplicated with AI additions, Oracle intends for Oracle Cloud Infrastructure (OCI) to facilitate AI adoption. This is a ‘more is more’ approach to customer choice.

Sicilia argued that only with dependable cloud infrastructure and the option for customers to choose their own AI model, data formats, and cloud residency, can AI success really be achieved.

“It’s no secret that at Oracle, we are building the world’s largest AI cloud,” Sicilia said.

“We think that this will all show up where the work actually happens, and that is servicing the AI directly in the applications and workflows that you're already living in, right at your fingertips. Whether you're in finance, HR, manufacturing, retail, healthcare, financial services and beyond.”

At Oracle AI World Tour London 2026, the firm unveiled 22 new Fusion Agentic Applications in line with this goal, powered by off-the-shelf AI agents trained in enterprise-specific tasks including identifying sales opportunities, coordinating supply chain operations, and automating cash flow.

We’ll have to wait and see if these agents deliver a real return on investment.

While data and fine-tuning play a major role in the success of AI deployment, Oracle sticks out as a hyperscaler without AI models of its own. Google Cloud offers its customers first-party Gemini offerings and Microsoft Azure boasts privileged access to OpenAI’s model catalog. AWS – which doesn’t spring to mind when I hear ‘frontier model developer’ – has put in work with its Nova model family, even if analysts question its confidence in the offering.

Oracle, by contrast, has no models of its own, nor an exclusivity deal with a major frontier lab. Though it was quick to form partnership with AI developer Cohere in 2023, whose models form the basis for Oracle NetSuite’s generative AI features, Oracle has subsequently bet on an agnostic approach.

Let’s get this out of the way: there’s obvious merit for the strength of OCI for running intense compute for enterprise AI. This is particularly when it comes to tasks such as training and fine-tuning LLMs, as Steve Miranda, EVP of Oracle Applications Development, explained to assembled media at the event.

“We have significant technical advantage in OCI, especially around the speed of network and security, which is why the Nvidia infrastructure runs best on OCI, which is why most, if not all, the top large language models in the world have chosen OCI to train their models,” said Miranda.

I suspect, however, that this won’t matter so much at the enterprise level. Business leaders aren’t generally looking to train or even fine-tune their own LLMs, they’re looking to deploy whichever pre-made AI tool delivers them the most value.

Customers willing to forgo flexibility and an open ecosystem in the name of the latest end-to-end AI deployment may find their head turned by Azure with its OpenAI tie ups, or the first-party model developer Google Cloud.

In the face of this competition, Oracle’s commitment to openness reads more like a bet against AI innovation, not for it. The offerings that Oracle thinks are make or break for customer retention aren’t AI at all, but Oracle Databases.

Indeed, Oracle has even gone to the length of partnering with other hyperscalers to make Oracle AI Database services available alongside competitor offerings. In these circumstances, customers can make use of OCI Exadata hardware directly next to, or even in, data centers run by AWS, Azure, and Google Cloud for low latency ease of access. This could be a major draw for customers looking for a low-friction, multi-cloud approach.

Nathan Thomas, SVP of product management, OCI at Oracle, explained how customer demand has driven OCI to support a varied, multi-cloud strategy.

“We hear from customers all the time that they want to use their AI pipelines in those environments,” he said.

“They want to be using Vertex, they want to be using Copilot, Sagemaker, Bedrock, and they want to have their data local to that, low latency, high performance, accessibility to it. But they still need to take all the governance with them.”

Over the long-term, Oracle is betting that customers will stick with it for both its openness and consistent, full-stack approach. Frankly, it’s too late to now change course – it can’t compete with the likes of Google Cloud for walled-garden AI capabilities and it seems to know this.

“There is nothing special, from a technical perspective, about the agents we build,” Miranda said, adding that the framework Oracle lets customers run its agents in a wide range of agentic frameworks.

Customers may choose stability, and neutrality, over specialization and the potential for lock-in when they choose their cloud provider. The uphill challenge for Oracle is convincing customers that this ‘nothing special’ approach is enough to give them an edge over competitors.

A simple and scalable foundation for enterprise data

Sicilia says one of Oracle’s major strengths is there’s “only one version of OCI”, allowing customers to plan for deployment without having to plan for compromises or region-specific versions.

“The form factor, the physical location, whether or not it's consumed in a multi-cloud environment, it doesn't matter,” he said.

“There's one version and that is, we think, to the customer’s advantage because we're able to look at all of the services, 100% of the services, no matter how big or how small the data center is, no matter how big or how small the consumption model is, as well as all the same privacy, security, cyber defenses.”

Thomas explained that Oracle’s approach allows customers to trust in a consistent application experience. He added that specializing in the creation of small clouds has fed directly into Oracle’s capacity for running private, distributed cloud regions directly in customer data centers, an offering he described as “a huge part of our sovereignty play”.

While digital sovereignty is a huge focus for hyperscalers, Oracle hopes its full-stack approach – as outlined by Sicilia – will help it stand out from the competition.

“This is an area where we really feel like a lot of the other cloud vendors have been let off the hook,” Thomas said.

“There's some sort of idea that, ‘hey, what OCI is doing is super special or different’. We kind of think that's backwards. You know, every cloud really should by default be able to scale down, like we've got all 150 OCI web services available on all of those form factors, the pricing is the same in all those environments, and it’s still metered billing the way that we have in any public region.”

Oracle is striving to meet rising demand for sovereign cloud in Europe amid geopolitical uncertainty and a push for reduced reliance on US cloud providers. The strength, in practice, of sovereign cloud controls compared to those offered by competing hyperscalers depends on the level of separation customers desire and unknowns such as the US CLOUD Act.

Like its competitors, Oracle argues that it offers ‘operator sovereignty’, allowing workloads segregated from the public cloud to be run not just within specific regions but by staff who are residents of that region. But this wouldn’t be enough to protect EMEA data from the US CLOUD Act, should Oracle choose to comply with a government directive to access data.

With geopolitics where it is, I suspect Oracle is making the argument a few years too late. European giants such as Airbus reportedly moving critical apps to a sovereign European cloud and White House PR surrounding Oracle becoming TikTok’s US secure cloud provider may have strengthened the association between the firm and the US government in the minds of EMEA customers.

It feels to me like sentiment toward US cloud providers across EMEA only goes in one direction right now, no matter how enticing the offer. In these conditions, Oracle may need to do more to lure new customers in and this may mean going beyond the slow and stable approach to which it’s currently committed.