Emerging markets are becoming a growing destination for data centre investment as technology companies race to build infrastructure for artificial intelligence, but the boom could expose weak power systems and widen development gaps unless governments invest alongside the industry, according to a July 2026 report.
Investment in data centres across emerging markets and developing economies rose from about $6 billion in 2015 to more than $25 billion in 2026, after reaching a record of about $31 billion in 2024, the report, Risks and Opportunities in Data Centre and AI Investment in Emerging Markets, said.
Globally, announced data centre investment surged to more than $320 billion in 2025, up 74% from a year earlier and representing almost a quarter of global foreign direct investment capital expenditure. Emerging markets and developing economies accounted for about 40% of announced projects, underscoring their growing role in the expansion of AI infrastructure.
The report, published by DEC Private Markets, says the investment race is increasingly being determined not simply by access to land or consumers but by the availability of reliable electricity, fibre networks, international connectivity, finance and predictable regulation.
That creates an opportunity for developing economies to attract capital while upgrading infrastructure but also exposes a critical vulnerability: AI workloads require substantially more power than conventional digital services.
Africa remains a smaller but emerging market
Asia dominates the data centre expansion among emerging markets, with China, Malaysia and Indonesia leading the growth. Brazil and Mexico have emerged as major Latin American hubs, while the United Arab Emirates is positioning itself as a regional AI infrastructure centre.
Sub-Saharan Africa remains much smaller in scale, with development constrained by power reliability and connectivity. South Africa, Kenya, Nigeria and Morocco are nevertheless attracting increasing investment.
Nigeria and South Africa have each attracted more than $5 billion in data-center investment announcements since 2020, according to the report. Africa has attracted more data centre FDI projects than emerging Europe despite persistent infrastructure constraints, although many of the projects remain at an early stage and could be delayed or halted.
Kenya illustrates the scale of investment beginning to reach the continent. Microsoft and G42 announced a $1 billion project in 2024 for a 1-gigawatt green data centre campus intended to host East Africa’s first Azure cloud region. The project is expected to support last-mile internet access for 20 million people in Kenya and 40 million across East Africa, alongside AI and digital-skills training.
In South Africa, Microsoft is investing about $300 million between 2025 and 2027 to expand hyperscale cloud and AI infrastructure, while Visa plans to invest $57 million through 2028 in its first African data centre to localise payment processing and improve latency and resilience.
Electricity becomes the new constraint
The biggest risk to the AI infrastructure boom is electricity.
Global data centre electricity consumption has been rising by roughly 12% a year since 2017, more than four times the growth rate of overall electricity consumption. Global data centre electricity use is projected to more than double to 946 terawatt-hours by 2030, with the US and China accounting for the largest increases.
For emerging markets with fragile grids, the implications are significant.
Hyperscale data centres operate continuously and require high-voltage electricity, redundant power systems and intensive cooling for GPU infrastructure. Without investment in substations, transmission and generation capacity, the report warns that additional AI demand could increase outages, raise electricity tariffs and intensify competition for reliable power between data centres, households and smaller businesses.
The problem is particularly acute for countries already struggling to meet existing electricity demand.
If renewable generation and grid expansion fail to keep pace, additional data centre demand could be met by coal and gas, potentially increasing fossil fuel dependence and complicating climate commitments. The report also identifies water as an emerging constraint because hyperscale facilities can require billions of litres of water annually for cooling in some cases.
The economic prize
The infrastructure requirements are substantial, but so are the potential economic gains.
The report says AI investment can raise productivity, accelerate digital adoption, develop workforce skills and generate new economic activity. Across 11 Asia-Pacific economies, AI spending and productivity spillovers were estimated to generate about $247 billion, equivalent to 1.6% of combined GDP in 2024, with most of the gains coming from productivity spillovers.
AI applications are already producing potential gains in agriculture, logistics and manufacturing by improving forecasting, reducing waste and streamlining supply chains. AI-enabled irrigation and fertilisation systems have been associated with yield increases of 5% to 15%, according to studies cited in the report.
Data centre projects can also generate activity beyond the facilities themselves. Construction creates demand for engineering and other services, while mature ecosystems can support software development, cybersecurity, cloud operations, data services and AI research.
Localised computing capacity could also reduce latency for fintech and logistics platforms, making real-time digital services more viable for local businesses.
But the employment effect is unlikely to be evenly distributed.
The report cites evidence that every 1% increase in AI spending raises demand for skilled labour by 0.21% while reducing demand for unskilled labour by 0.15%. That creates a risk of greater wage polarisation if education and skills development fail to keep pace with technology adoption.
Investment follows infrastructure
The emerging-market competition is, therefore, becoming a race to build an ecosystem rather than simply attract a data-centre campus.
The report identifies reliable power, backbone fibre, international connectivity, available land, permitting, financing and predictable data-governance rules as key determinants of where investment flows. Sovereign AI strategies and data-residency requirements are also increasingly encouraging countries to host computing and data domestically.
Malaysia, for example, is using its semiconductor base and energy-transition strategy to position itself as an AI infrastructure hub. India, Thailand and other Asian economies are similarly combining cloud investment with national AI strategies and skills programmes.
The scale of some commitments illustrates the stakes. Google has announced $15 billion for an AI data centre and hub in India, while Amazon Web Services has committed $12.7 billion to expand Indian cloud regions. Malaysia has attracted commitments from Google, Microsoft, AWS and Nvidia-linked investors running into billions of dollars.
For countries with weaker infrastructure, however, the report points to a potentially different path.
As AI shifts from intensive model training toward real-time inference, smaller AI models and edge computing could allow emerging economies to capture some benefits without building the enormous hyperscale infrastructure required for frontier-model training.
Lightweight models running closer to users could use existing mobile networks and reduce dependence on centralised computing facilities, creating a potential route for countries constrained by electricity grids and limited data centre capacity.
Policy will determine who captures the boom
The report’s central message is that data centres alone will not guarantee economic transformation.
Infrastructure provides access to AI, but productivity gains depend on whether businesses, households and governments can actually adopt the technology. That requires affordable connectivity, digital skills, complementary investment and predictable regulation.
Countries that coordinate digital expansion with energy planning and grid modernisation are more likely to become preferred destinations for hyperscalers, the report says. Malaysia, Brazil and the UAE are cited as examples of economies aligning power strategies with AI infrastructure ambitions.
For African economies, the investment opportunity is, therefore, larger than the data centres themselves. The ability to secure reliable power, expand fibre and international connectivity, develop skilled workers and create credible regulatory frameworks could determine whether the continent becomes a meaningful participant in the AI economy or remains primarily a consumer of computing capacity built elsewhere.
The report’s warning is ultimately an economic one: AI infrastructure can become a development catalyst, but only if the electricity, connectivity, skills and institutions needed to turn computing capacity into productivity are built at the same time.
