World Bank flags a ‘Four Cs’ roadmap to close the AI divide: usage surges in middle-income nations, but compute and skills gaps persist

## Morocco’s AI moment, seen through the World Bank’s ‘Four Cs’ lens On November 21, 2025, the World Bank published a new factsheet. It summarizes early findings from its Digital Progress and Trends Report 2025. The message is clear. AI is spreading fast across developing economies, but foundations are uneven. The note describes a dual reality. Usage and job demand are rising across middle-income markets. Yet models, startups, funding, and data-center power remain concentrated in rich economies. This post applies that lens to Morocco and outlines a practical path. ### Key takeaways - Morocco can harvest quick gains with “small AI” while building the Four Cs. - Compute and skills gaps are real, but demand signals are strong. - Local language, open data, and sector pilots unlock near-term value. - Public-private compute access can accelerate safe experimentation. - Train people, upgrade infrastructure, and measure outcomes rigorously. ## Momentum is real, and Morocco can ride it By mid-2025, more than 40% of ChatGPT’s global traffic came from middle-income countries. Brazil, India, Indonesia, and Vietnam led that wave. Employers’ demand is also surging. Global GenAI vacancies grew ninefold from 2021 to 2024, and one in five postings are in middle-income markets. These signals matter for Morocco. They show that skills can be built and exported from middle-income contexts. They also point to growing remote work opportunities. Morocco can prepare talent and capture higher-value digital work. ## Concentration remains steep, which shapes Morocco’s constraints High-income countries account for 87% of notable AI models and 86% of AI startups. They host 91% of AI venture capital while holding just 17% of the world’s population. They also host 77% of global co-location data-center capacity. Upper-middle-income countries hold 18%, lower-middle 5%, and low-income under 0.1%. That concentration affects access and adaptation for lower-middle-income economies. Compute, funding, and research networks cluster abroad. The World Bank highlights open-source AI as a counterweight. It helps democratize participation and localization, including for language and sector needs. ## Small AI first: a practical playbook for Morocco The factsheet urges a “small AI” approach now. Focus on lightweight, locally relevant applications that run on everyday devices or modest cloud. Improve clinical decision support, help small firms market products, and upgrade public-service delivery. This is a pragmatic way to leapfrog constraints while larger investments mature. For Morocco, this approach fits the current mix of assets and gaps. Smartphones are widespread, and cloud access exists. But affordable high-performance compute is limited. Targeted, efficient models can deliver value under present conditions. ## The Four Cs for Morocco: what to build and how ### 1) Connectivity: reliable, affordable internet and power The Bank stresses access gaps. Usage reaches about 93% of people in high-income countries. It drops to 54% in lower-middle-income and 27% in low-income economies. Morocco needs dependable power and internet in schools, clinics, and municipal offices. Businesses also need affordable bandwidth to use AI services. Practical moves: - Prioritize power reliability and backup for digital hubs, clinics, and schools. - Extend fiber to industrial zones and university campuses with shared backhaul. - Support community networks and public Wi‑Fi in rural service centers. - Lower device and data costs through targeted vouchers and SME bundles. ### 2) Compute: affordable access to cloud, devices, and local data centers Middle- and low-income countries together hold only around 23% of global co-location capacity. Morocco can combine local colocation, public cloud, and shared GPU access. The goal is predictable, affordable, and secure compute for startups, researchers, and agencies. Open-source models can stretch limited resources. Practical moves: - Establish a public‑private compute access program with clear eligibility and quotas. - Procure shared GPU nodes in local data centers with research and startup lanes. - Negotiate cloud credits and discounted egress with major providers for national programs. - Pool demand across agencies to lower unit costs and simplify procurement. The World Bank highlights similar tools in other contexts. Its materials describe open data frameworks, public‑private compute access, and language‑technology programs. They also reference country pilots such as Nigeria’s national compute initiative and a multilingual LLM effort. Morocco can adapt these patterns to local needs and regulations. ### 3) Context: languages, governance, and content for national needs AI works best when tuned to local language and data. Morocco’s daily language mix includes Arabic, Darija, Amazigh, and French. Public services, commerce, and tourism often require multilingual interactions. High‑quality, consented datasets are the cornerstone. Practical moves: - Fund creation of open, privacy‑compliant corpora for Arabic, Darija, and Amazigh. - Launch a language resource consortium with universities and media partners. - Update open data inventories with machine‑readable, well‑documented datasets. - Build domain ontologies for health, agriculture, justice, and tourism to improve model tuning. Data governance is essential. Morocco’s data protection regime can guide compliant data sharing and model training. Public agencies should adopt data minimization and purpose limitation. Clear licensing improves reusability and trust. ### 4) Competency: skills from digital literacy to advanced AI engineering Skills are the binding constraint. Fewer than 5% of people in low-income countries have basic digital skills. Roughly 66% do in high-income economies, the Bank notes. Morocco needs breadth and depth. Civil servants need practical AI literacy. Engineers need hands‑on model training and MLOps skills. Practical moves: - Roll out short, job‑aligned AI literacy for teachers, clinicians, and administrators. - Provide micro‑credentials in data analysis, prompt engineering, and model evaluation. - Expand applied AI programs in universities and coding schools, with industry projects. - Support apprenticeships with local firms and remote work placements with global teams. ## Moroccan use cases that fit constraints and deliver value Agriculture: AI can support irrigation scheduling, pest alerts, and yield forecasts. Satellite and drone imagery can flag water stress for olives, citrus, and cereals. Farmers can receive recommendations on mobile devices. Cooperatives can optimize logistics and reduce spoilage. Health: Lightweight decision support can help triage common conditions in primary care. Pharmacy inventory tools can predict stock‑outs and suggest procurement schedules. Referral systems can prioritize cases based on risk signals. Messaging bots can deliver maternal health reminders and appointment nudges. SMEs: Generative tools can draft product descriptions and ads in Arabic, French, and English. Assistants can create invoices and answer common customer questions. Sales teams can segment customers from transaction data. Exporters can translate catalogs and compliance documents quickly. Tourism: Multilingual chatbots can answer itinerary questions and safety policies. Voice assistants can help visitors navigate transit and payments. Hotels can automate routine queries and upsell services. Guides can use translation aids in crowded sites. Public services: Document summarization can speed case reviews. Chatbots can triage citizen requests before human handoff. Customs can use risk scoring to focus inspections. Municipalities can classify and route complaints from multiple channels. Education: Teachers can use grading assistants for short answers and essays. Reading tools can support learners in Arabic, Darija, and Amazigh. Administrators can automate repetitive forms. Students can explore safe, curriculum‑aligned tutoring tools. Logistics and trade: Morocco hosts major trade infrastructure, including Tanger Med. AI can optimize truck flow and yard operations. It can forecast container dwell times and mitigate congestion. Small pilots can start with narrow, explainable models. ## Startups, campuses, and public agencies are key actors Morocco’s startup scene includes teams working on applied AI. Companies such as ATLAN Space showcase AI‑enabled autonomy for environmental monitoring. Health ventures like DataPathology illustrate digital diagnostics ambitions. Such firms can anchor pilot projects and mentor new talent. Universities and coding schools are expanding data and AI coursework. They already partner with industry on applied research. Public agencies are digitizing workflows and records. These foundations support “small AI” pilots with measurable outcomes. Technoparks and incubators can coordinate cross‑sector cohorts. Priority sectors include agriculture, tourism, logistics, and health. Each cohort should access shared datasets, compute, and mentors. The goal is to ship working prototypes within months, not years. ## Governance and safety: earn trust from the start The Bank emphasizes “no‑regrets” investments and safe adoption. Morocco can pair pilots with practical guardrails. Start with clear data protection, role‑based access, and audit trails. Add model cards, risk assessments, and human‑in‑the‑loop designs. Procurement should require transparency on data sources and model limits. Vendors should provide evaluation metrics and failure cases. Agencies should test for bias and language coverage. Sunset clauses can ensure low‑performing tools are retired quickly. ## Funding and sequencing: make the near term count Invest where benefits are broad. Power, broadband, devices, data, and skills have economy‑wide returns. That is the Bank’s “no‑regrets” guidance. Morocco can align national programs and donor support around these priorities. A practical 12‑month plan: - Refresh open data portals with quality targets and API access. - Launch a Small AI Challenge Fund focused on public service and SME tools. - Stand up a shared GPU pool in local colocation with fair‑use policies. - Secure multi‑year cloud credits for startups and universities. - Fund a national language corpus program for Arabic, Darija, and Amazigh. - Train 10,000 learners through short, job‑aligned micro‑credentials. - Pilot AI assistants in three agencies with external evaluation. A practical 24–36 month plan: - Scale proven pilots across ministries and regions. - Expand compute capacity and interconnect university hubs. - Establish a national model evaluation lab and testbed. - Introduce sector sandboxes for health, finance, and transport. ## Metrics that matter Track progress with simple, transparent indicators. Examples include: - Connectivity uptime at schools, clinics, and municipal offices. - GPU hours delivered to researchers and startups at subsidized rates. - Open datasets added, with documentation and update frequency. - Number of language resources released under open licenses. - Share of civil servants and teachers trained with assessed proficiency. - Pilot adoption rates, error rates, and citizen satisfaction scores. ## Bottom line The World Bank’s message is pragmatic. Diffusion is outpacing foundational capacity in many places. Closing gaps in connectivity, compute, context, and competency unlocks inclusive gains now. Morocco can move fast on “small AI” while building toward larger opportunities. The recipe is actionable. Invest in power and broadband. Expand shared compute and open data. Build language resources and skills at scale. Pilot targeted tools, measure results, and iterate. Morocco does not need to wait for perfect infrastructure. It can deliver useful, safe systems today. The Four Cs provide a clear map. The time to execute is now.