In Rabat, the official launch of “AI Made in Morocco” on January 12 put on the table a structured, sequenced 2025–2030 roadmap designed to move Artificial Intelligence (AI) from intention to administration, infrastructure, and the real economy. The document sets out a clear order of operations. First, build trust. Then strengthen the technical backbone that digital services depend on. Next, organize skills development, place-based innovation, and financing, before judging the whole effort on measurable results.
Behind the title, the approach is unambiguous. Morocco is not treating artificial intelligence as a plug-in for existing systems, but as a national worksite that forces simultaneous changes in law, infrastructure, data, skills, and governance. The roadmap says so plainly, relying on ten programs that cover the full cycle, from regulation to change management. This choice matters, because AI never “works” on its own. It depends on system quality, available data, security, teams capable of maintaining it over time, and an administration able to absorb change.
2025–2026, trust as the starting point
The first phase, spanning 2025–2026, places trust at the center. The state plans a “Digital Law X.0” and an “AI Framework”, followed by a certification process and a trust label starting in 2027, and alignment with international frameworks by 2029–2030.
This timeline is not a rhetorical exercise. As soon as AI touches public services, it engages administrative responsibility and citizen protection. It requires clarity on what is permitted, what must be controlled, how errors are handled, and what level of transparency is expected. The plan’s logic is to reduce uncertainty from the outset, so sensitive uses are not deployed on a legally fragile basis.
Infrastructure before promises
This regulatory framing goes hand in hand with a technical project that conditions almost everything else. The roadmap explicitly places the “Move to Cloud” program in 2025–2027. The message is clear. Before promising intelligent assistants and AI-augmented services, the country needs an architecture that can absorb the load, secure access, ensure continuity, and govern environments.
In the same spirit, the plan sets heavy infrastructure milestones, including a 50 MW data center and the expansion of the Benguérir data center, then, toward the end of the period, a target of a 500 MW “Green DC” mega-campus. These figures point to a material reality. Beyond software, AI is also an industry of compute and hosting, and therefore an investment, energy, resilience, and competitiveness issue.
Data and interoperability, the nerve center of reform
The most structuring section, because it touches the administration’s internal functioning, lies in the “platform of digital commons”. The roadmap announces a first version of a software forge and a library of twenty critical APIs, then the large-scale structuring of public data pools in Open Data, before moving toward interoperability and a generalized “once-only” principle.
This last point may sound technical, but it is one of the most political. It means no longer asking citizens and businesses for information that an administration already holds. That is where the most concrete reform plays out, the reform of everyday public service. And that is also where AI’s basic condition lies. An “intelligent” system simplifies nothing if data is fragmented, reference datasets are inconsistent, or administrations do not talk to one another.
Training at scale, and above all retaining skills
On human capital, the plan stands out for a degree of detail that is rarely made public. It provides for the creation of coding schools, including YouCode, a digital school, as well as an online platform for certified training. It also cites the “Master AI Junior” program launched in October 2025, and the Royal Moroccan Football Federation (FRMF) “young talents” program intended to train more than 200,000 beneficiaries aged 8 to 18. The roadmap also sets volumes, with 2,500 trained as early as 2025, then a target of 14,000 trained cumulatively, alongside an upskilling program, JobInTech, aimed at building practical digital skills to improve employment prospects.
It also strengthens the system of Doctoral Teaching Assistants in digital fields (Doctorants-Moniteurs, DM), with quantified steps, 150 in 2025, then 200 in 2026 and 2027, before stabilization. The whole picture suggests a two-speed strategy, a broad base to feed the market and administration, and a scientific reinforcement to produce applied research and high-end expertise. The obstacle is the same everywhere. Training is essential, but stabilizing teams and retaining experienced profiles is just as critical. It is a question of attractiveness, career paths, and the ability to keep talent working sustainably within the public sector.
The LLM project, the bet on a Moroccan model
The roadmap also embraces a symbolic and potentially decisive project, the one around large language models (LLMs). It plans an alpha version of the national model in Darija and Amazigh, then large-scale training and integration into public services, followed by export ambitions by 2029–2030. The language choice aligns with the public-utility objective. An AI that does not understand users’ language remains peripheral.
But this pillar will be judged on proof. Measurable quality, security, robustness, data respect, and the ability to integrate into real public-service journeys. This is where a strategy can generate tangible gains, or collide with technical and organizational limits.
Innovation beyond Rabat, the promise of territories
Another ambition plays out outside Rabat, in the regions. The roadmap provides for deploying the national network of Jazari Institutes across all twelve regions, developing a “Digital Lab Ed Tech”, and ultimately reaching financial autonomy for the institutes. It also sets an operational milestone, a “Jazari Root” for steering and execution.
Territorializing innovation is a direct answer to the risk of a transformation concentrated in a few hubs. But it only makes sense if it rests on clear missions, teams, budgets, and a pipeline of projects. Without that, a network becomes a label. With it, it can become a tool for spreading skills, structuring local ecosystems, and producing solutions tailored to regional realities.
Financing and the emergence of national champions
To move beyond a prototype economy, the plan finally tackles financing and champion-building. It mentions venture building and venture capital mechanisms, the creation of a venture fund, then the scaling-up of incubation and acceleration instruments, leading to the emergence of national tech champions. Here again, the issue is not to multiply startups, but to produce companies able to sell, hold markets, scale, and export. Along that path, public procurement, data availability, purchasing speed, and regulatory clarity become levers as decisive as capital.
A technology diplomacy with an African dimension
International cooperation is also positioned as a standalone axis through the Hub Morocco Digital for Sustainable Development (D4SD), a pan Arab-African center of excellence in artificial intelligence and data science launched in September 2025.
Building on that momentum, the plan provides for South-South technology projects and regional partnerships, as well as an objective to organize a World AI Summit in Morocco. It is a way to place the Moroccan strategy within a technology diplomacy with an African dimension, without giving up international visibility.
The real test, execution
The most demanding step remains the one that always makes the difference between a roadmap and tangible transformation, implementation. To get there, the plan announces a new organizational chart with a Directorate-General for AI, data-driven steering, and dynamic adjustments to the roadmap. It then provides for measuring socio-economic impact.
It also puts adoption explicitly on the agenda, with “digital relays”, technology demystification campaigns, responsible adoption by state agents and SMEs, and finally a final evaluation of socio-economic spillovers. This is an implicit recognition of a simple fact. AI does not impose itself by decree. It spreads through uses, trust, training, and the capacity to integrate new tools without destabilizing services.
Concrete milestones, a promise verifiable from 2027
The 2025–2030 roadmap therefore tells the story of a framed and highly structured ambition. It does not promise a technological miracle, but organizes a transformation built on concrete prerequisites, cloud, data centers, data, interoperability, skills, and governance. Above all, it places the country in front of verifiable deadlines.
By 2027, the markers will be straightforward to observe. The cloud shift, the existence and use of critical APIs, the quality of Open Data, and real interoperability capacity. From there, AI can become a tool for reform and competitiveness. If those conditions are not met, it will remain a set of projects, interesting but dispersed.
