NVIDIA’s Marc Domenech on Saudi Arabia’s shift from AI ambition to deployed compute
NVIDIA’s VP for Enterprise across the Middle East, Türkiye, Africa and Southern Europe, talks about Saudi Arabia’s move from AI announcements to deployed compute, the energy question, and where GCC demand goes next
14 September, 2026
TT
16
At the recently concluded LEAP 2026, Saudi Arabia’s AI story shifted from announcements to hardware on the ground. The first phase of HUMAIN’s build-out is now underway on NVIDIA’s Blackwell Ultra infrastructure, the opening move in a plan to develop AI factories with up to 500 megawatts of capacity and several hundred thousand GPUs over five years.
Marc Domenech, NVIDIA‘s VP for Enterprise across the Middle East, Türkiye, Africa and Southern Europe, talks to Gulf Business about what’s now operational in the kingdom, why energy efficiency, not just raw power, will decide who wins, and how the Saudi and UAE ecosystems will evolve as the region moves from building AI models to running them at scale.
LEAP 2026 marked a shift from AI investment announcements towards actual deployed capacity in Saudi Arabia. What is now operational, and how quickly do you expect utilisation and demand for AI compute to grow from here?
LEAP demonstrated that Saudi Arabia is moving decisively from ambition to execution. The first phase of HUMAIN’s build-out is now underway, centred around NVIDIA Blackwell Ultra infrastructure designed to support advanced model development and large-scale inference in the kingdom.
This is the beginning of a much broader roadmap. NVIDIA and HUMAIN announced plans to develop AI factories with capacity of up to 500 megawatts over five years, supported by several hundred thousand NVIDIA GPUs. That capacity will scale progressively as facilities, energy, networking, software and customer requirements develop together.
The next measure of progress is utilisation. Training advanced models requires substantial computing power, but inference, using those models to serve people and businesses, creates continuous demand. As organisations move beyond pilots and deploy AI agents and applications across their operations, that demand will become broader and more sustained.
Saudi Arabia has the investment, energy resources, talent ambitions and industrial base to move quickly. Success will ultimately be measured not by installed capacity alone, but by what that capacity produces: locally relevant models, new applications, scientific advances, more productive industries and globally competitive companies.
The NVIDIA–HUMAIN collaboration envisages AI factories with up to 500 megawatts of capacity and several hundred thousand NVIDIA GPUs over five years. What will infrastructure at that scale enable Saudi Arabia to do?
It will give Saudi Arabia the computing foundation to develop and deploy AI at national and industrial scale.
An AI factory is different from a traditional data centre. A conventional data centre stores information and runs applications; an AI factory turns data into intelligence. It brings together accelerated computing, high-performance networking and software to train models, adapt them using local knowledge and operate them at scale.
This level of capacity can support Arabic-language models, AI agents, scientific research, industrial digital twins, robotics, autonomous systems, and more. It can also enable organisations to work with complex datasets and develop applications tailored to the kingdom’s priorities.
Shared infrastructure can lower the barrier to innovation. A startup, university or business should not need to build its own supercomputer before it can use advanced AI. Access through cloud services can give a much wider ecosystem the computing and software required to build, test and scale new ideas.
The hardware itself is not the outcome. Its value will be determined by what grows around it: developers creating products, startups building companies, researchers advancing science and established industries becoming more productive. That is how computing capacity translates into durable economic value.
As Saudi Arabia builds local AI infrastructure, how important is it that the kingdom develops its own models, data capabilities and technical expertise rather than simply importing computing power and technology?
It is critical. Computing capacity provides the foundation, but lasting value comes from combining it with local data, models, software, talent and industry expertise.
Every country has its own language, culture, institutions and economic priorities. Saudi Arabia therefore has a clear opportunity to develop AI that reflects its context, from Arabic-language models to applications designed for sectors such as energy, healthcare, financial services, logistics and manufacturing.
This does not mean developing every layer of the technology independently. AI advances through global research, common technology platforms, open models and international collaboration. The opportunity is to use that global foundation to build capabilities that are locally relevant, locally operated and aligned with the kingdom’s priorities.
Talent is what connects infrastructure to outcomes. Developers need the skills to build and optimise models. Enterprises need people who can deploy AI securely and reliably in production. Researchers and startups need access to computing and software so they can turn ideas into working applications.
The countries that derive the greatest value from AI will not necessarily be those that own every element of the technology stack. They will be those that can turn their knowledge, data and expertise into intelligence that improves services, strengthens industries and creates new opportunities.
Power, cooling and access to energy are becoming major constraints on AI data-centre expansion globally. Could energy availability eventually become a bigger bottleneck than access to GPUs?
Energy will be a defining consideration for AI infrastructure everywhere. The central question, however, is not simply how much power is available, but how efficiently that power can be converted into useful intelligence.
That requires treating the AI factory as one integrated system. Computing, networking, cooling, software and the facility itself must be designed together. Accelerated computing is fundamental because it performs AI and high-performance computing workloads far more efficiently than general-purpose architectures.
Rack-scale design, high-speed networking and direct liquid cooling can improve performance and computing density. Software is equally important: optimised models and inference engines can reduce the resources needed to produce each result.
The Gulf has an opportunity to design new facilities around these requirements from the outset, rather than adapting data centres built for an earlier generation of computing. The region also brings extensive experience in developing and operating large, complex energy systems.
Energy availability will remain an important part of every infrastructure decision. The industry’s responsibility is to keep improving efficiency across chips, systems, networking, cooling and software. The objective is to produce more intelligence—and greater economic value—from every watt.
NVIDIA is also working on skills development, robotics, physical AI and digital twins in Saudi Arabia. Where do you expect the first meaningful commercial applications outside the technology sector?
The earliest applications are likely to emerge in sectors where Saudi Arabia already has deep expertise, substantial physical assets and clearly defined operational challenges.
Energy and manufacturing are strong examples. Companies can use digital twins to simulate facilities, production lines and industrial processes before making changes in the physical world. This can improve design, maintenance, worker safety and operational efficiency.
In logistics, AI can optimise warehouses, ports and distribution networks, while autonomous systems and robotics can support repetitive, complex or physically demanding work. Construction and infrastructure operators can simulate projects, test different scenarios and identify potential problems earlier.
Healthcare and life sciences also present significant opportunities, including medical imaging, genomics, drug discovery and AI assistants that help researchers and healthcare professionals work with complex information. These applications must be developed with the appropriate safeguards and specialist expertise.
Physical AI takes this further by enabling machines to perceive, reason and act in the real world. Training and testing those systems in physically accurate simulations before deployment can shorten development cycles while reducing cost and risk.
Adoption will move fastest where AI addresses a measurable need. The strongest projects will begin with the desired outcome: higher productivity, greater safety or better service, and apply the right technology to achieve it.
Saudi Arabia and the UAE are both making ambitious investments in AI. How do you see their respective ecosystems evolving, and where will the strongest demand for NVIDIA technology emerge across the GCC over the next three to five years?
Saudi Arabia and the UAE are each building on different economic strengths and institutional capabilities. We do not view their progress as a race with a single winner. Growth in either market can strengthen the wider region by attracting talent, investment and companies to the GCC.
Saudi Arabia has an opportunity to apply AI across industries operating at significant scale, including energy, manufacturing, logistics, healthcare and major infrastructure projects. Its investments in computing capacity and skills can support economic diversification and the development of locally relevant services.
The UAE has established itself as an international business and technology hub, supported by strong research institutions, cloud providers, technology companies and global connectivity. Its ecosystem is advancing across infrastructure, research, enterprise adoption and digital services.
Over the next three to five years, demand across both markets will increasingly shift from building models to running them. AI agents will transform knowledge work, digital twins and robotics will reshape physical industries, and Arabic-language applications will serve people and organisations throughout the region. This growth in inference will create continuous demand for efficient, high-performance computing.
Across the wider GCC, the opportunity extends beyond GPUs. Customers will need complete AI infrastructure: computing systems, networking, software, models and technical expertise. NVIDIA’s role is to provide that full-stack platform and work with the ecosystem to move AI from experimentation into reliable, large-scale production.






















