AI Investment Boom Leaves South Africa's Job Market Exposed to Disruption

Dario Amodei, chief executive of Anthropic, has put a number on it: half of entry-level white-collar jobs could vanish within five years. That projection, whether or not it proves accurate, is already reshaping how capital is allocated globally, and South Africa sits in an uncomfortable position relative to both the investment flows and the disruption they carry.

Data-centre spending has become the primary growth engine for the US economy and a significant driver for China’s expansion. The concentration of capital in AI infrastructure signals how thoroughly the technology has captured investor attention. For countries far from the centres of technological development, the question of who captures AI’s economic value is not abstract. It is structural.

Amodei’s scenario is stark. A narrow elite of capital owners, he argues, will control the platforms, compute power, and electricity required to run advanced AI systems. Agentic software displaces white-collar workers; humanoid robots substitute for blue-collar labour. Most people are left holding few assets and offering labour with negligible economic value.

Scepticism is warranted, though. These projections come from executives whose company valuations depend on investor belief in transformative impact. Labour-market data so far offers scant evidence of widespread disruption. Daron Acemoglu, a Nobel Prize-winning economist, estimates total factor productivity gains from AI at under 1% over a decade, concentrated in narrow sectors. The gap between industry rhetoric and measured economic effect remains wide.

Historical patterns of innovation distribution are instructive here. A study of early-2000s US patents found that for every dollar of value created by invention, roughly 30 cents accrued to workers while 70 cents became profit distributed to shareholders. If AI breaks the assumption of stable labour and capital output shares, returns could concentrate further in capital holders’ hands. Displaced cognitive workers would likely crowd into occupations machines have not yet disrupted, suppressing wages across sectors nominally considered AI-safe, including construction, social care, and logistics.

South Africa’s exposure operates on multiple fronts. Business process services, call centres and back-office operations in Cape Town and Durban, represent the country’s most successful job-creating export industry of the past decade, with more than 70,000 jobs created in the Western Cape alone. Large language models handle voice and text processing competently at negligible marginal cost, threatening this sector’s viability directly. While India faces larger absolute exposure, South Africa’s concentration in this industry makes the risk proportionally more severe.

By contrast, the country’s capital markets offer virtually no hedge against that disruption. Naspers and Amsterdam-listed Prosus, leveraging Tencent investments, represent the principal AI-related exposure on the JSE. A national hedge dependent on a single stock and a Chinese platform provides minimal protection. Compounding the problem, South Africa’s electricity constraints fundamentally limit data-centre development, the infrastructure foundation for AI operations. The country cannot easily participate in the capital-intensive infrastructure buildout driving global growth elsewhere.

South Africa already contains a permanent economically marginalised population. Roughly one-third of the labour force faces unemployment on narrow definitions, exceeding 40% on broader measures, with youth unemployment running substantially higher. The relevant economic question is not whether AI creates a bifurcated society but whether it will widen existing divisions, and whether South African democracy can absorb the consequential socioeconomic pressures that follow.

Investor behaviour itself signals uncertainty about AI’s actual productivity effects. Many investors acquire AI exposure as a hedge rather than from conviction about profitability. If AI transforms everything and you lack exposure, you lose entirely. That is not a forecast. It is an acknowledgment of fundamental uncertainty, and uncertainty presents very different challenges for those holding assets than for those without them. Whether South Africa can build meaningful equity participation in the sector before the infrastructure window narrows is the question that will define its economic position in the decade ahead.