Sama's Kenya Layoffs After Meta Contract Loss: The Labor Economics of AI's Invisible Workforce

On 16 April 2026, TechCabal reported that Sama — the San Francisco-headquartered AI data services company that markets itself on an ethical outsourcing model and employs workers in Kenya, Uganda, and India to perform data labeling, content moderation, and AI training work — would lay off more than 1,100 Kenyan workers following the end of its Meta contract. The news was described, in TechCabal's framing, as dealing "a fresh blow to Kenya's fast-growing but fragile AI outsourcing sector." That framing is accurate in describing the economic injury but insufficiently precise about its structural causes. What happened at Sama is not an event that requires a narrative of fragility; it requires a structural analysis of the terms on which Global South labor is integrated into the AI development economy and the power asymmetries that those terms encode.

The Sama-Meta relationship illustrates precisely what Rana Foroohar means when she distinguishes between makers and takers in the contemporary economy. Meta's platform — Facebook, Instagram, WhatsApp — generates tens of billions of dollars in quarterly revenue from advertising sold against content whose safety, quality, and relevance is maintained, in substantial part, by low-wage African and Asian workers performing content moderation and data labeling. That labor is structurally invisible in Meta's public accounts, outsourced through intermediaries like Sama, and therefore outside the firm's legal obligations on labor standards, benefits, severance, and job security. John Kenneth Galbraith's concept of countervailing power — the idea that concentrated corporate power generates offsetting institutional power in the form of unions, regulatory agencies, and civic organizations — has no adequate expression in this context. The 1,100 Kenyan workers who lost their jobs when Meta chose not to renew had no collective bargaining mechanism, no portable benefits, and no enforceable claims on the platform corporation whose training data their labor produced.

The Anatomy of AI Labor Extraction in East Africa

The data-labeling economy that employs workers across Kenya, Uganda, Ethiopia, Ghana, and increasingly across South Asia functions on what researchers studying platform labor have described as a piecework model for the digital age. Workers are typically employed by intermediary firms — Sama, Appen, Scale AI, CloudFactory, and similar organizations — under contracts that specify deliverables rather than hours, establishing an employment relationship that is legally more analogous to contractor status than to direct employment while in practice involving structured working hours, supervised workplaces, and mandatory software tools.

Sama has historically differentiated itself within this ecosystem by claiming that its model provides higher wages, better benefits, and more ethical working conditions than competitors. That claim, while partially substantiated by wage comparisons with informal sector alternatives in Nairobi, should not obscure the structural arrangement underlying it. Sama's ability to offer any working conditions — "ethical" or otherwise — is entirely dependent on the continuation of its contracts with platform corporations that retain complete discretion over whether to renew, expand, or terminate those contracts based on their own commercial considerations, including cost reduction, automation of labeling tasks through AI models that themselves required earlier human labeling to train, and geographic diversification of their supply base.

The Meta contract that Sama lost had involved content moderation work for Facebook and Instagram — review of graphic violence, hate speech, sexual content, and other policy violations — as well as structured data annotation for AI training purposes. Content moderation, unlike some forms of data labeling, is not easily automated: it requires human cultural and linguistic judgment that AI models cannot yet reliably replicate across the full range of contexts and languages in which Meta's platforms operate. The workers who performed this function were not, in any meaningful sense, in a declining industry. They were in a structurally vulnerable labor market position because the firm to which their employer was subcontracted retained all the negotiating power.

Kenya's AI Outsourcing Sector and the Structural Limits of Digital Development

Kenya's government and development agencies have, over the past decade, made substantial investments in positioning Nairobi as a hub for digital economy employment. The Konza Technology City project, the licensing of data center operators, the expansion of undersea cable connectivity through the TEAMS and EASSy cables — these are all elements of a deliberate strategy to insert Kenyan labor into global technology supply chains. The theory of development implicit in this strategy is that participation in digital labor markets, even at the bottom of value chains, generates skills, wages, and institutional capacity that support upward movement over time.

The Sama layoffs expose the structural limits of this theory without necessarily refuting it entirely. The problem is not that Kenya's workers are unqualified or unproductive; the workers who lost their jobs at Sama were, by all available accounts, performing their functions competently. The problem is that the value chain they occupied is structured in a way that concentrates all of the pricing power and all of the contract discretion at the platform corporation level while distributing all of the employment risk downward to the workers at the bottom. analysts of the entrepreneurial state would note:

Michael Porter's competitive advantage framework would note that Kenya's current position in the AI services value chain — providing undifferentiated, easily substitutable data labeling and content moderation labor — is inherently precarious. The path to sustainable competitive advantage would involve moving up the value chain into more complex AI services, model evaluation, domain-specific annotation requiring specialized knowledge, and eventually AI development itself. Nigeria's Enugu State, according to a separate April 2026 report, is explicitly pursuing this trajectory with plans for an AI institute oriented toward talent export in higher-value categories. Whether this trajectory is achievable at the speed necessary to absorb labor displaced by platform contract terminations is an open question.

The Power Asymmetry and What Regulation Cannot Yet Reach

The structural issue at the heart of the Sama-Meta-Kenya triangle is a jurisdictional gap in labor protection that operates across the three nodes of the relationship. Sama is incorporated in California and subject to US labor law with respect to its direct employees but not, in any enforceable sense, with respect to the obligations it owes workers it employs through subsidiary entities in Kenya. Meta is similarly incorporated in the United States and subject to US employment law for its direct workforce, but its contractual relationship with Sama is a commercial relationship between corporations, not an employment relationship — meaning that Meta bears no legal obligation to Sama's Kenyan workers even though the work those workers perform is central to Meta's product.

Kenyan labor law applies to Sama's operations in Kenya but is limited in its effective reach by enforcement capacity, by the political sensitivity of imposing requirements that might deter foreign investment, and by the fundamental asymmetry of a negotiation in which the employer can credibly threaten to move the contract to another country. This is, in 's terms, a structural feature of the international division of labor in the current hegemonic cycle: capital retains spatial mobility while labor does not, and regulatory jurisdiction follows national boundaries that capital can cross freely.

The workers who lost their jobs at Sama are not a constituency that features prominently in the Washington conversations currently underway about AI regulation, labor displacement, and the governance of platform corporations. The IMF gathering in Washington this week — overshadowed, according to The Guardian, by the Iran war's energy price shock — will not produce policy frameworks for protecting data-labeling workers in Nairobi from platform corporations' contract decisions. The V&A, closer to home for UK readers, was simultaneously facing calls from its own workers for living wage status as it opened its Stratford site. The pattern is consistent: the labor that enables the AI economy is systematically undervalued, poorly protected, and concentrated in the jurisdictions with the least regulatory capacity to address that pattern.

Monexus reported this as a labor-power and global value-chain story rather than a local employment event because the structural conditions that produced Sama's layoffs will produce identical outcomes wherever Global South AI labor is integrated into platform corporation supply chains without enforceable upward obligations.