The Madrid stop of the Oracle AI World Tour 2026 It had little of a “regional” event. The presence of Mike Siciliaglobal CEO of the company, turned the day into a kind of declaration of intent about the role that Spain and Europe are called to play in Oracle’s Artificial Intelligence strategy. Beyond the usual deployment of demos and use cases, the message was clear: AI stops being an experiment and becomes a business operating system that rests, above all, on the data quality and in the sovereignty of the infrastructure that supports it.
In the opening keynote, after the introduction of Albert Triolacountry leader of Oracle Spain, it was Mike Sicilia who carried all the weight of the speech, who did not hesitate to say that “in the more than 30 years that I have been involved in technology, I have never seen a technology advance as quickly and efficiently as AI has done”, although he related that speed to very specific architecture, data governance and consumption model decisions.
The data above all
One of Mike Sicilia’s most relevant statements was the explicit shift towards inference on private data as the main economic driver of enterprise AI. As he explained, many of the great language models are already trained on Oracle Cloud Infrastructure (OCI), but the real value is in how organizations exploit their own data for inference.
This inference, he defended, cannot be separated from the data plane or the infrastructure where the most sensitive loads reside. Hence the commitment to AI Data Platform from Oracle, which is presented as a data vectorization layer as a service, designed to index and serve semantic contexts regardless of where the databases or data lakes are deployed: on-premises, in other public clouds or in hybrid environments.
The message that Mike Sicilia left for those present at the keynote was that the discussion should no longer focus on “which model to use”, but on “how to prepare my data so that any model can work on it, without breaking sovereignty or existing security controls”.
“Not everything has to be in the Oracle stack”
Another important nuance, far from the classic vendor lock-in discourse, was his insistence that Oracle’s value in AI does not involve forcing all data to be concentrated in its technological platform. “Sometimes it is thought that to get value from Oracle AI, everything has to be in the Oracle stack, and that is simply not true,” said Mike Sicilia. The proposal involves working with partners and also with competitors, in an approach where the data platform and vectorization act as interoperable glue on top of different clouds and data engines.
This connects with what Albert Triola later reinforced in other forums at the event itself: Oracle does not want to be, at least not exclusively, a “model provider”, but rather the data and governance platform on which clients can consume the LLMs of OpenAI, Google, Meta or others, depending on their use case and compliance requirements.
Spain as a sovereign cloud laboratory
The strategic burden of the Oracle AI World Tour 2026 in Madrid cannot be understood without what happened the day before the event. Mike Sicilia held an audience with the King Felipe VI and a work meeting with Oscar LopezMinister for Digital Transformation and Public Service. The Government itself highlighted that this was the first visit by the CEO of Oracle to a European country since his appointment, in a context in which Spain wants to consolidate itself as a pole of attraction for digital investments and public-private collaboration projects around the sovereign cloud and AI.
In this context, Mike Sicilia recalled that three Oracle cloud regions already operate in Spainand that a good part of the design of the EU Sovereign Cloud It has been inspired precisely by conversations with clients and with Spanish administrations – both central and regional. From these discussions emerged the current model of dual-level sovereignty: on the one hand, the eurozone as a common regulatory framework; on the other, country-specific deployments and controls that respond to public sector, banking or critical sector requirements.
In Madrid, Minister Óscar López’s speech reinforced that narrative: European funds have been used, as he explained, as a “factory for public-private collaboration projects” that encourage companies like Oracle to multiply their investments in the country, relying on the sovereign cloud and AI as levers to modernize the administration and improve the competitiveness of the business fabric.
From network digital twin to GPUs as a service
The presence of Spanish clients on stage sought precisely to give credibility to that speech. My SpengerCEO of The Orangesdetailed how the operator is deploying more than 150 AI use cases, with a special focus on network automation and customer experience. One of the most striking projects is the construction of a digital twin of the mobile network in the Oracle cloud, on which to simulate changes and optimize capacity before touching the productive environment.
Meini Spenger also announced that MasOrange is working with Oracle to offer GPU capacity as a service to the Spanish business fabric, converting the infrastructure that currently supports its own AI cases into a marketable asset for third parties. In practice, this positions the operator as an “aggregator” of accelerated computing capacity on OCI, at a time when the availability of GPUs is one of the most serious bottlenecks to scaling generative AI projects.
For its part, Mar Donategeneral director of Bimbo Group Spainfocused on the less glamorous but most transformative part of AI: the fine optimization of operations, logistics and supply chain. The combined use of advanced analytics and AI integrated into Oracle applications is allowing them visibility over demand, reducing shrinkage and better adjusting production, an area in which thousandths in forecasting translate into margin points.
The advantage of the “full stack” over generalist hyperscalers
On a competitive level, Mike Sicilia was especially forceful. For Oracle, the differentiating element is not so much in having “a better model” than other hyperscalers, but in being able to orchestrate the complete stack: infrastructure, database, semantic data layer, SaaS applications and analytics under the same architecture and governance model.
He stressed that most cloud competitors do not play “at scale” in either business applications or transactional databases, two domains in which Oracle has been operating for decades. From that position, the argument is that time-to-value in AI is drastically reduced when the provider can encapsulate “invisible complexity” in turnkey services that expose AI capabilities directly within business processes, without forcing customers to set up an AI factory from scratch.
In that sense, the speech links with the idea already present at Oracle AI World 2025 that Oracle’s AI is “built in, not bolted on”, that is, it is not about adding a “copilot” on top of an existing ERP or HCM, but rather that these applications, from the beginning, are capable of operating with agents, recommendations and automations that understand the business context in real time.
From pilots to operational AI
One of the keys that hovered over the day, both in the keynote and in the round table with Oracle vice presidents Binoy Sukumaran (Database Cloud), Jason Rees (Technology and Cloud Infrastructure EMEA), Glenn Beeswanger (Applications y SaaS Engineering) y Rich Clayton (Product Strategy for AI and Analytics), was the need to move from the pilot to mass deployment. Several Oracle executives have already dubbed 2026 “the year of AI operationalization.” After a 2025 full of proofs of concept, many organizations have been blocked in the scaling phase due to lack of operating model, talent or data governance.
Oracle’s thesis is that only a full-stack approach, with a strong data governance layer, can break that lock. For those present at the event, the implicit message was twofold: on the one hand, that the debate is no longer “whether” to adopt AI, but “how” to do it without multiplying risks of information leakage, shadow AI or excessive dependence on a single model provider; on the other hand, that the alignment between infrastructure, data and applications is not only a system architecture preference, but a requirement to be able to demonstrate traceability, explainability and compliance in regulated environments.
Spain, key piece in the Oracle model
That all this has been staged in Madrid is no coincidence. Spain combines a public sector immersed in an accelerated digital transformation with large, highly internationalized corporations in banking, telecommunications, food or industry. It is, in practice, an ideal testbed for the type of AI that Oracle wants to sell: intensive in its own data, highly regulated, distributed across multiple clouds and extremely sensitive to European sovereignty and compliance.
The audience with the King and the meeting with Minister Óscar López send an additional signal: the conversation about AI, sovereign cloud and vectorized data has already left the technical perimeter to be placed at the center of the country’s industrial and digital policy. For Oracle, Madrid is not just another stop on the AI World Tour, it is a key piece in how the company wants to articulate its presence in the EU and demonstrate that its promise of “trust, choice and scale” can be landed in real environments, with names and surnames.
Leaving the event, the feeling is that the battle of enterprise AI is not going to be decided solely by who has the largest model or the best benchmark, but by who manages to be closest to the data (and the authorities that regulate it) without breaking the agility that businesses need. In Madrid, Oracle has made it clear where it wants to position itself in that equation.
The entry Mike Sicilia, CEO of Oracle, places data sovereignty and quality at the center of the AI strategy appears first in MuyComputerPRO.
