For more than three decades, Jensen Huang was the undisputed face of the AI hardware boom — a leather-jacketed kingmaker whose nod at a conference could validate a startup’s entire roadmap. That era came to an abrupt halt this week when Huang resigned as Nvidia’s CEO, capping a tumultuous fortnight dominated by revelations of multiple extramarital affairs — including one with an executive US intelligence reportedly suspected of espionage ties to Beijing — alongside a federal investigation and mounting congressional scrutiny. With CFO Colette Kress stepping in as interim chief and Nvidia’s board scrambling to chart a future without its co-founder, partners across the global AI ecosystem are suddenly asking what comes next.

Yet just weeks before the scandal upended Silicon Valley, Huang was doing what he had long done best: working the room at GTC Berlin, interrogating founders on unit economics and signing off on promising edge hardware. Among those in front of him was Mohamed Benouda, founder and chairman of Moroccan tech group ABA Technology — by his own account, the only African partner in the building.

Following a first encounter in Paris in 2025, Benouda had returned to show off the finished product: Fusion AI In-a-Box, an on-premise system designed to slash deployment costs and keep sensitive industrial data out of foreign clouds. Huang not only quizzed the Casablanca team on their cost-per-token metrics and urged them to go global, he also took a marker to their machine, leaving a simple dedication: “To ABA.”

In a candid interview with Launch Base Africa — portions of which were first published by Moroccan outlet Le360 on 13 November — Benouda looks back on those encounters and explains why Morocco’s push for sovereign AI infrastructure is built to outlast any single Silicon Valley leader.

How did you first meet Jensen Huang?

I first met Jensen Huang on 11 June 2025, at an evening hosted by NVIDIA during GTC Paris. NVIDIA had brought together its international technology partners to share its direction and discuss ongoing projects.

The atmosphere was very relaxed. Jensen spoke informally, with his knack for making complex technology accessible. I was in the front row. At the end of his talk, I went up to him and we discussed the Fusion Computing System we were developing at ABA Technology.

Our relationship with NVIDIA’s teams had started the year before. In 2024, we presented this system to Deepu Talla, Vice President and General Manager of Robotics and Edge Computing. Our idea was to bring different computing capabilities (GPU, CPU and NPU) together in a single box, so that organisations could deploy their own AI solutions, including using open models, while keeping control of their data and their operations.

Deepu liked the concept very much and encouraged us to keep working with his teams, in particular Davide and Charbel. So when I approached Jensen in Paris, I could talk to him about a project that was already moving forward with his team.

What was your first impression of him as a person?

A smiling, cheerful and very courteous man. That is what struck me immediately. Here is someone who plays a major role in the world’s technological transformation, and yet the conversation stays simple and natural.

His vision is impressive, of course. But when you meet someone in person, it is also the small things you remember: the way he welcomes you, listens to what you present and engages in the conversation.

That first evening left me with the memory of an approachable leader who clearly still enjoys talking about technology and what can be built with it.

You met him a second time recently. What happened?

The second meeting took place at GTC Berlin, on 20 October 2026. In the meantime, our project had evolved a great deal. We had officially launched Fusion AI In-a-Box in April 2026, with two complementary components: a Factory, to turn business needs into qualified, deployable AI solutions, and an Execution platform, to run these solutions end to end, from data to applications, through models, agents and workflows.

This time, we could show him the concrete result of our work.

At NVIDIA, the technological bar is very high. To pursue a collaboration and contribute to designing projects, you have to move forward, demonstrate and bring something. So presenting our work there mattered a great deal to our teams.

My strongest memory is the moment Jensen signed our box, with the dedication “To ABA”. For those of us who had carried the project from its earliest days, it was a proud and emotional moment.

What struck you most in that exchange?

His questions. He wanted to understand the features, the market we were addressing and how we had brought AI software and computing infrastructure together in a single system.

I felt he particularly appreciated our drive to industrialise the making of AI solutions: shortening the time it takes to go from a business need to an operational application, but also gaining better control over running costs, in particular the cost per token.

What struck me was how he went straight from the technology to its usefulness and its market. He looked at what we had built, but also at what we could do with it on a larger scale.

He encouraged us to aim for the global market. We talked in particular about our positioning in sovereign AI, which he saw as a direction with a real future. Receiving that encouragement at this point in our journey meant a lot to us.

Did you talk about Morocco?

Yes, and it was an important subject for me. We told him that, to our knowledge, we were the only African partner there.

I spoke to him about Moroccan talent and the transformation our country is undergoing under the leadership of His Majesty King Mohammed VI. I mentioned the automotive and aerospace industries and infrastructure, such as Tanger Med, and the fact that we now have a Ministry of AI and that His Majesty has singled out this sector as critical and strategic for the country: examples that show Morocco’s ability to build industrial and digital ecosystems and take its place in global value chains.

The conversation turned to an ambition close to my heart: making Morocco a true AI nation, able to play a leading role in the region.

That requires several things to come together: training and mobilising talent, developing computing capacity, growing national champions and enabling a startup ecosystem to build its own solutions. Morocco can develop skills across the entire AI value chain, from data and compute to models, agents and the execution of applications.

That is the ambition I wanted to convey to him.

What role did Jensen Huang play in your relationship with NVIDIA?

The relationship was built over time with NVIDIA’s teams. They played an important role from the moment we presented our concept in 2024, and our work with his team then allowed us to continue development.

My meetings with Jensen gave this story a personal dimension. In Paris, I presented our approach to him. In Berlin, I could show him what it had become.

His signature remains a wonderful memory. But what I remember above all is the conversation about the product, the questions he asked us and his encouragement to take our ambition international. For a Moroccan team building its own technology, these are moments that count.

How could Jensen Huang’s departure affect your collaboration with NVIDIA?

Jensen’s encouragement remains a powerful memory. Our collaboration also rests on relationships built since 2024 with NVIDIA teams across EMEA and the United States, strengthened by our work in mining, healthcare, industry and public services.

With Fusion AI, we aim to cut time to production by a factor of up to seven and AI project costs by a factor of three to four, depending on the deployment. This ability to deliver results is what underpins our long-term collaboration.

Could his departure change the outlook for sovereign AI, particularly in Africa?

The need for sovereignty extends beyond any individual. A mine, a hospital or a government agency must retain control over its data and ensure operational continuity.

Consider predictive maintenance at an African mine: anticipating the failure of critical equipment allows teams to intervene before production stops. As an illustration, avoiding 100 hours of downtime at $10,000 per hour represents $1 million in avoided production losses, before deducting the costs of the solution and maintenance interventions.

Sovereign AI enables industrial data to be processed on site, reduces dependence on external connectivity and preserves know-how. For industrial companies, the challenge is to build this capability and retain the value it creates.