Global insurers and reinsurers with clients in the compute, data centre and power utility sectors are in for a tumultuous decade as demand grows for innovative risk transfer solutions to protect increasingly complex and costly infrastructure.
A Goldman Sachs Global Institute scenario analysis, drawing on Goldman Sachs Global Investment Research and Nvidia projections, estimates around US$7.6 trillion in cumulative artificial intelligence (AI) capital expenditure between 2026 and 2031. As the stock of insurable assets grows, so does the demand for insurance.
Although not in the same lofty territory as the United States (US), Europe or Asia-Pacific, Africa looks certain to benefit from the AI-driven infrastructure theme as investors seek greater geographic diversification. The concentration and natural catastrophe risks emerging in established data centre markets reinforce that case.
Swiss Re estimates that around 40% of US data centre capacity is likely to be located in areas with a significant to very high annual frequency of tornadoes, while about a quarter will be in areas frequently exposed to large hail.
McKinsey & Company says that demand for Africa-based data centre capacity will swell from around 0.4 gigawatts (GW) in 2025 to as much as 2.2GW by 2030, requiring up to US$20 billion in new investment. Many of these data centres require power and water infrastructure upgrades, creating additional opportunities.
New data centres
In its 2026 outlook report, the African Energy Chamber estimates that South African data centres will consume more than 16 TWh of electricity annually by 2030, equivalent to around 6.5% of the country’s projected electricity consumption that year.
Several promising continent-wide projects had been announced or were advancing by mid-July 2026, including a US$400 million investment in a data centre expansion programme in Egypt; a US$400 million, 20MW data centre powered by a 100MW on-site gas plant in Ogun State, Nigeria; another US$250 million for a 100MW data centre in Lagos, Nigeria; and Teraco’s 40MW JB7 facility in Johannesburg, South Africa, as part of a broader expansion programme financed through an R8 billion (US$490 million) syndicated loan.
In related news, Equinix recently announced plans to expand its South African data centre footprint. The group said it had acquired R890 million (US$55.5 million) worth of land in Johannesburg and Cape Town as part of a broader R7.5 billion (US$467.9 million) investment programme aimed at adding 160MW of data centre capacity and capitalising on growing demand for cloud and AI infrastructure.
In July 2026, the Cape Town Municipal Planning Tribunal gave preliminary approval to a land-use, subdivision and consolidation application that could pave the way for two large data centres at King Air Industria, next to Cape Town International Airport. If built, these Equinix facilities would cover more than 120,000 square metres and have a combined electrical demand of around 174MW. Site development and building plans have not yet been considered, while details on backup power, emissions and water consumption were not clear from the land-use application.
Africa’s ability to attract compute and data centre investments hinges on access to reliable and scalable power. Matthew Turner, head of innovation at SolarAfrica, believes that the continent’s abundant solar and wind resources can be leveraged to support AI-related infrastructure growth and shape a more sustainable model for powering it.
“As AI becomes more autonomous, the energy systems supporting it will need to become more innovative and flexible to enhance its abilities,” he wrote in a recent thought leadership piece.
Opportunity for insurers
Insurers and reinsurers will be eying the premium on offer as the aforementioned data centre and similar projects get off the ground. During the ‘Sigma 2/2026 World Insurance in 2026’ media launch, Swiss Re estimated that cumulative global data centre property and casualty insurance premiums could reach US$90 billion between 2024 and 2030, while up to US$400 billion in capital could be required to support these projects.
Commenting at the media launch, Ivan Gonzalez, CEO of Swiss Re Corporate Solutions, warned of rising accumulation risks as the AI capex theme unfolds.
“Some of these data centres are highly concentrated in a geography,” he said. Risk managers were encouraged to use data centres as a case study for the interconnected risks their businesses and shareholders face.
Gonzalez described projects in the AI compute, data centre and energy disciplines as both an opportunity for the insurance industry to show how it contributes to resilience and an underwriting challenge.
“Traditional underwriting is no longer sufficient; you now need to understand the engineering design, the operational resilience and the accumulation risk,” he said.
The developed market numbers thrown about during this global media conference will leave many African insurance stakeholders in awe, with some data centres pencilled in at around US$20 billion in asset values alone, while also carrying sizeable business interruption exposure.
Gonzalez conceded that the asset value for some of these projects was “a bit mind-boggling” before even considering operational risks such as cooling, electricity or network failures, alongside cyber exposure.
Whether or not these projects can find sufficient insurance support will likely hinge on the markets they are located in.
“If you have a big data centre project being constructed somewhere in Southeast Asia, the ability of industry to give you a traditional placement might be there,” Gonzalez said. “But sometimes [capacity] might be limited in terms of the scope [or] inadequate in terms of the overall framework”. In such cases, “the importance of innovative, or what we call alternative risk transfer solutions”, cannot be overlooked.
Global outlook
Zooming back out to the global view, Sigma 2/2026 identified supply shocks, fragmentation in capital markets and AI-driven infrastructure as the main trends affecting insurers and reinsurers. The report noted that the Middle East conflict that dominated the first half of 2026 was the fourth major supply shock to the global economy in the past six years, and that such shocks were typically inflationary, with a negative impact on GDP growth.
According to Jérôme Haegeli, chief economist at Swiss Re Institute, the reinsurer had lowered its global real GDP growth forecast by 0.3 percentage points, to a still-robust 2.5% in 2026, and raised its inflation outlook by 1.0 percentage point to average 4.0% for this year. On the plus side, the world was described as “entering a period of large-scale investment in physical capital and infrastructure, driven by AI adoption, electrification, the energy transition and industrial policy”.
Swiss Re said that total global insurance premium growth would slow to about 1.3% in real terms in 2026, with slightly better performances from the life sector and emerging markets excluding China.
The non-life sector was described as being in a cyclical slowdown, but forecast return on equity of 11.4% in 2026 remained well ahead of the cost of capital. AI capex was held up as providing positive cross-currents for growth, with the five hyperscalers projected to spend US$750 billion in 2026 alone.
Interconnected risk puzzle
The natural catastrophe exposure mentioned earlier in this article is just one piece in this interconnected risk puzzle. According to Gonzalez, data centres have significant interdependencies on electricity supply, sophisticated cooling systems and increasingly connected operational technologies.
Underwriting decision-makers will have to consider these interdependencies alongside construction and repair lead times. “Supply chain risk is critical,” Gonzalez said, after observing that the components for a US-built data centre may have to be sourced from over 90 countries.
One of the closing concerns addressed during the Sigma 2/2026 media launch was whether insurers, reinsurers and other allocators of capital might be left high and dry if the valuation ‘bubble’ that had developed around US technology shares were to burst. Haegeli responded by saying that the capital flowing to AI opportunities could hardly be seen as speculative in light of the current high-interest rate environment.
He argued that “even if AI stocks were to take a large hit and move 20% or 30% lower, the capex would remain in place”.
Haegeli concluded that AI-related expenditure in compute, data centres and power infrastructure was underpinned by long-term structural demand and multi-year investment commitments, rather than movements in the stock market or individual share prices.
COMMENTS