AI Supplier Concentration and Corporate Dependency: The Future of Work and Domestic Value Capture in Egypt

Artificial intelligence (AI) reaches firms through a chain of interdependent layers: specialized hardware, cloud and data-center infrastructure, data, foundation models, application programming interfaces (APIs), and business applications (Gambacorta & Shreeti, 2025). Market power can arise at different points in this chain, and the evidence is mixed. Large, fixed costs, economies of scale, and scarce inputs favor concentration in some segments (Korinek & Vipra, 2024), while the OECD (2026) reports dynamic entry in some model segments alongside persistent structural risks in hardware, data, and cloud infrastructure. Describing the whole AI economy as a single monopoly is therefore not defensible without specifying the market, layer, geography, and period. 2. Concentration in a supplier market is also not the same as dependency in a particular firm, and neither is the same as employer power over workers. Supplier concentration describes how market shares are distributed. A firm's structural exposure describes how much weight one supplier relationship carries in its operations. Dependency arises when that exposure combines with limited ability to replace the supplier. Employer labormarket power concerns the relationship between firms and workers (Azar & Marinescu, 2024). Linking these three is a theoretical possibility, not an empirical regularity. Athey and Scott Morton (2025) model imported AI as a priced input whose upstream pricing can affect downstream costs, wages, and welfare, but they present a mechanism, not evidence that every AI-importing economy experiences it. Evidence on adoption points in several directions: AI investment has been associated with firm growth and employment gains (Babina et al., 2024), and exposure indices measure what AI could do to tasks, not realized job loss (Gmyrek et al., 2025). 3. Egypt is a useful setting for examining these links. The OECD (2024) review identifies constraints on access to computing infrastructure and data, especially for startups and small and medium-sized enterprises (SMEs), and records concern about vendor lock-in while local capacity is built. The National AI Strategy for 2025–2030 sets objectives for scalable infrastructure and cloud services, data accessibility, startups and SMEs, and local talent (National Council for Artificial Intelligence, 2025). Existing Egyptian labor-market research measures task automatability from online job postings (Dawoud et al., 2026a, 2026b), but postings do not show which suppliers employers use or how exposed firms are to them. Dependency on external providers is also well studied in IT-outsourcing research (Ngah et al., 2024), so the contribution here is not the idea of dependency itself but its application to AI-specific mechanisms, such as model replacement, API integration, and changing service conditions. 4. This article therefore asks three questions. First, what does the available evidence show about concentration and contestability across layers of the AI value chain? Second, what do official and academic sources show about Egypt's AI infrastructure, labor-market exposure, and digital-services base, and where do they conflict? Third, what links between supplier exposure, operational flexibility, workplace deployment, and domestic value retention are plausible, and which can only be tested with firm-level data? It answers them through an integrative review of academic and institutional literature and a secondary analysis of publicly reported indicators. Neither method measures firm-level exposure, so the article proposes an analytical framework and identifies evidence gaps. It does not estimate causal effects or claim to have measured dependency. 5. Section 2 sets out the review and indicator methods. Section 3 reviews market structure across the value chain. Section 4 examines Egypt's infrastructure, labor-market, and policy context. Section 5 develops the framework and the conditions under which supplier exposure could matter for workers. Section 6 discusses policy implications for contestability, interoperability, and domestic capability, and Section 7 sets out limitations and a research agenda.