Romain Bottier, SE Lead, META, VAST Data, discusses the changing demands of AI infrastructure as organizations across the Gulf and beyond look to expand their AI capabilities. 

[For more news, click here]

Artificial intelligence is changing the demands placed on the infrastructure behind modern computing. As organizations move beyond experimentation and begin integrating AI into their operations, questions around data, computing capacity and the systems connecting them are becoming increasingly important. The challenge is no longer simply adopting new technology, but building the foundations needed to make it work at scale. 

For Romain Bottier, SE Lead, META, VAST Data, these changes intersect with years of experience in high-performance computing and AI infrastructure.  

The Gulf has become an important market in this shift, with the UAE and Saudi Arabia stepping up investment in technology and digital infrastructure. Their efforts reflect a wider change across the industry, as businesses and cloud providers face growing demands on the systems that support their AI operations. 

In this exclusive interview with Tech Revolt at AI Everything Abu Dhabi 2026, Bottier discusses the changing technology landscape and the forces shaping the next phase of AI development. 

Building Sovereign AI Beyond the Data Center 

Q1: Having watched the Gulf’s AI market evolve since VAST entered the region in 2021, what do you think the UAE and Saudi Arabia are uniquely positioned to build that could give them an edge in the global AI economy? 

The UAE and Saudi Arabia have different strategies, but the overall vision is similar. They want to bring AI and technology into their countries and make them available to their people. There is also a very strong focus on sovereignty. They don't want to be overly dependent on the US cloud hyperscalers. They want to control their own destiny when it comes to artificial intelligence. 

At the government level, we can see a real drive and a genuine willingness to advance AI for the benefit of the country. It's not about doing it purely for the sake of profit; it's about helping people optimize, automate and simplify tasks, and enabling businesses to be more productive and grow. They also have the funding and financial capability to make it happen. 

I think Saudi Arabia and the UAE are very well positioned to achieve this because they understand what they can do with AI. They have a clear vision of what's possible and how to get there. 

Q2: You’ve argued that sovereignty is about control, not simply residency. As models, data and infrastructure increasingly come from different providers, where does maintaining that control become most difficult? 

The topic of sovereignty has gained a lot of traction over the past year. When people think about sovereignty, the first thing that comes to mind is data location. Where is my data? Is my data leaving my country, crossing my borders or moving beyond my control? 

But the conversation has advanced significantly over the past 12 months. Residency remains important, but sovereignty increasingly depends on the conditions under which data can be accessed, and how that access is governed and audited. Organizations need to know not only where their data resides, but also what happens to it and who can interact with it. 

That becomes more complicated when you introduce AI models. Some are open, while others are proprietary technologies that their developers are understandably reluctant to expose. At the same time, organizations want to use the most capable models available without surrendering control of their sensitive data. Both sides have something valuable they need to protect. 

That's the challenge we're addressing with VAST DataEnclave. It allows sensitive enterprise data and proprietary AI models to come together within a hardware-isolated, confidential computing environment. Cryptographic attestation verifies that the environment meets the required security conditions before encryption keys are released, allowing the model to process the data without either party surrendering control of its underlying assets. 

The result is greater choice. Organizations can access advanced AI capabilities while retaining control over their data, and model providers can make their technology available without exposing their intellectual property. 

Q3: VAST operates at the data layer of AI. As organizations move from conventional enterprise computing to AI-native environments, what fundamentally changes in how data needs to be handled? 

Legacy environments were not built for AI. They were built for specific tasks, whether that meant a shared file system for enterprise use or a transactional database. Different systems served different purposes, and these systems operated as silos. 

The AI narrative has changed the discussion. You now need to deal with every type of data, whether it's structured, unstructured or simply flowing around as streams. What you need, therefore, is a single platform that can ingest all these different forms of data and transform them to make them relevant. 

That can happen through data analytics, because the VAST Data platform can also be used as a data lakehouse. Alternatively, we can make the data available through AI pipelines for AI use cases. 

This applies whether you're building standard AI inference practices or moving toward agentic AI, where we can make transformed data available through MCP access. 

Q4: The Gulf is making major investments in GPU capacity. Beyond acquiring the GPUs themselves, what needs to be in place to ensure that compute is actually being used efficiently and translated into production AI? 

Buying GPUs alone is not necessarily what drives AI forward. In fact, this is a problem we've seen in the market. Everybody has been rushing to buy the technology, but the question is: how do you use it? You need a stack. You need to build a stack that can support use cases on top of it. 

We cover a big portion of that stack, with a strong focus on data. We provide the ability to move data in and out, stream data, and transform it into structured data that artificial intelligence can use and leverage. 

That was the missing piece for a lot of customers. They would buy GPUs and fast storage, but there was nothing in between to do something meaningful with the data that the GPUs could easily leverage. We facilitate that entire data layer, and we work with different technologies to help enable use cases. That's what's needed. 

As GPU capacity becomes more widely available, differentiation will increasingly come from the value around the compute, so data services, orchestration, software, developer experience, and the ability to help customers get to production. Neoclouds offering GPUs and a bit of storage are evolving to offer more.  

It shouldn't be a race to the bottom. It should be a race to value. 

The Commercial Stakes of AI Infrastructure 

As artificial intelligence becomes more established across industries, the commercial stakes are changing. Businesses are under pressure to identify where the technology can deliver meaningful returns, while technology providers face growing expectations to offer more than computing capacity alone. 

That shift will influence competition across the industry, from established infrastructure companies to a growing field of cloud providers. The companies best placed to succeed will be those that can demonstrate practical value to customers, rather than relying on the scale of their technology offerings alone. 

For business leaders, the question is no longer just whether to invest in AI, but how to make those investments count. 

Related Articles 

AI Is Turning Data Sovereignty Into an Enterprise Infrastructure Problem 

Saudi Arabia Is Building the Infrastructure Behind Its AI Ambitions at LEAP 2026 

Rethinking GPU Cloud Infrastructure as Inference Workloads Rapidly Expand