ITCEQ Analysis Note No. 78, published in May 2026 and authored by Olfa Bouzaiene, examines the integration of artificial intelligence within Tunisian companies through a statistical lens.
The study is based on the concept of absorptive capacity theorized by Cohen and Levinthal in 1990, structured around four dimensions: acquisition, assimilation, transformation and exploitation of technologies.
Drawing on the survey “Tunisian companies in the era of digital transformation,” the report identifies five business profiles and proposes three AI adoption trajectories.
The study’s central finding is striking. While 86% of Tunisian companies view new technologies as a driver of innovation, only 19% say they have actually innovated, and just 11% make use of public research and development incentives (ITCEQ, 2026, p.9).
This gap between intention and implementation lies at the heart of the issue, yet it is precisely here that the study’s reasoning begins to show its limits.
The diagnosis relies entirely on self-reported data. When the report states that 76.2% of private companies have a website and 72.5% use ERP-type management systems, or enterprise resource planning software (ITCEQ, 2026, p.6), it measures declared equipment rather than actual usage.
A company may own an ERP system without truly using it, maintain an outdated website, or classify a simple WhatsApp group as a “collaborative platform.”
The methodology does not distinguish between ownership and effective use, weakening the entire statistical framework built upon it.
Another issue lies in the decision to group AI, Big Data, Cloud computing and Blockchain into a single category, justified by a “logic of technological disruption threshold” (ITCEQ, 2026, p.15).
These technologies operate according to fundamentally different dynamics. Using cloud services is mainly an outsourcing decision, whereas deploying AI requires internal analytical capabilities and structured data.
The report itself acknowledges that its nonlinear principal component analysis captures only 46% of total variance in the first two dimensions (ITCEQ, 2026, p.18), meaning that more than half of the information escapes the conclusions presented.
The study’s static nature creates another difficulty. The report proposes three AI adoption trajectories (ITCEQ, 2026, pp.21–22): a sequential approach, direct acceleration and a complementary logic, even though it is based on a snapshot taken at a single point in time. No longitudinal data documents the actual paths followed by companies.
These trajectories are therefore theoretical constructs projected onto static observations, which weakens their prescriptive value. Even more problematic is the absence of any international comparison.
Are the 19% of companies that innovated performing poorly, moderately or reasonably for a middle-income economy? Without benchmarks from Morocco, Egypt, Turkey or Senegal, readers lack any standard against which to assess Tunisia’s performance. The report presents its figures in isolation, making them difficult to interpret politically.
The study’s disconnect from Tunisia’s economic reality also raises questions. The report identifies insufficient financial resources as the main obstacle, cited by 70.9% of companies (ITCEQ, 2026, p.9), without examining the structural conditions behind that shortage.
The cost of capital in Tunisia, restrictions on access to foreign currency for subscribing to foreign cloud services, and the scarcity of local venture capital are entirely absent from the analysis.
The same applies to the shortage of digital skills, cited by 63.3% of companies, which is treated solely as a training supply issue without any mention of the brain drain of Tunisian talent toward Europe and the Gulf. Training more professionals without addressing retention is like filling a leaking bucket.
The study also reflects an implicit bias in favor of technological adoption. It never questions the assumption that every company should digitalize, even though for some sectors or business sizes, non-adoption may be an economically rational choice.
Digital transformation is treated as a universal imperative, which is neither self-evident nor demonstrated.
Finally, the report’s recommendations (ITCEQ, 2026, pp.23–24) remain strikingly generic. Auditing needs, investing in monitoring systems and modernizing information systems are recommendations that could apply to almost any emerging economy.
No specifically Tunisian lever is identified, whether it concerns the potential of the country’s tech diaspora, the possible role of the free zones of Bizerte and Zarzis, or South-South partnerships with French-speaking African markets, which nonetheless represent a natural strategic horizon for Tunisia.
The report ultimately provides a carefully crafted statistical snapshot, but it stops halfway between diagnosis and actionable policy proposals.
