The $200 Billion Policy: Swiss Re Sizes the Insurance Bill of the AI Capex Boom

Every September, the reinsurance industry descends on Monte Carlo for the Rendez-Vous de Septembre, and every September one document ends up in every underwriter’s bag: Swiss Re Institute’s newest sigma report. This year’s edition, released on 5 September under the title Time to build: Expanding the frontier of insurability for the capex super-cycle, arrives with a headline built for the AI era: the global investment boom in AI data centres and renewable energy could generate around USD 200 billion in cumulative commercial insurance premiums between 2026 and 2030.

That is a striking number for an industry that spent the last decade worrying about shrinking commercial lines. It is also, Swiss Re warns, a number that comes with a catch — the same boom is manufacturing risk concentrations the insurance market has never had to price before.

The capex super-cycle, in numbers

The sigma report frames the moment bluntly: the global economy has entered a capital expenditure super-cycle, and AI infrastructure sits at its centre. Global energy investment is expected to reach USD 3.4 trillion in 2026, with roughly USD 2.2 trillion flowing to renewables, nuclear, grids, storage, low-emissions fuels, efficiency and electrification. The five largest US hyperscalers alone are expected to plough nearly USD 800 billion into AI-related capital expenditure this year, while estimates for global data-centre capex now exceed USD 1 trillion.

All of that capital needs insurance — during construction, commissioning, and the decades of operation that follow. Swiss Re Institute puts the resulting premium opportunity for AI data centres and renewable energy infrastructure at roughly USD 200 billion cumulatively through 2030.

For context, this builds on the Institute’s earlier work: in March 2026 it estimated that premiums tied specifically to data centres would grow from around USD 10.6 billion to USD 24.2 billion per year by 2030. The new USD 200 billion figure widens the lens to the full AI-plus-energy infrastructure complex over five years — a measure of just how fast the addressable market has expanded in six months.

From IT asset to strategic infrastructure

The core of Swiss Re’s argument is a category shift. Data centres are no longer information-technology assets tucked inside office parks; they are strategic infrastructure whose power requirements are measured in gigawatts, whose asset values are counted in the billions, and whose operations hinge on electricity grids, telecommunications, cooling systems and cloud infrastructure all working at once.

That shift cuts both ways. It creates deep, multi-line insurance demand — property, business interruption, construction, liability, cyber-adjacent covers. But it also means a single physical event can now cascade across policyholders, industries and lines of business simultaneously.

The report identifies four structural drivers of risk accumulation:

1. Increasingly large individual assets. Some AI data-centre campuses, including their computing equipment, would cost up to USD 50 billion to replace — an insured value on a single site that rivals the balance sheet capacity of entire national insurance markets.
2. Geographic clustering. Data centres gravitate to where power, land, water and connectivity are cheap and abundant. Texas and Virginia alone account for more than 40% of current and planned US data-centre capacity.
3. Supply-chain dependencies. Critical equipment such as high-voltage transformers now carry lead times of multiple years, stretching out project delays and business-interruption losses long after the physical damage is repaired.
4. Shared physical and digital networks. Common electricity grids and fibre routes can transmit disruption to otherwise unrelated businesses.

The catastrophe map nobody designed

Swiss Re’s natural-catastrophe math is where the report turns uncomfortable. More than a quarter of US data-centre capacity sits in areas that could see at least three days of large hail per year; around 40% is exposed to at least three tornado days annually. Texas — a top-two host state — sits squarely in both zones.

The Asia mirror is starker: some 88% of Taiwan’s semiconductor fabrication plants are located in extreme to very extreme seismic-risk zones. Given Taiwan’s centrality to global chip supply, a major quake there would not just be a property loss; it would propagate downstream through every industry that depends on advanced silicon — including the very AI build-out driving the premium boom.

And in a detail that will resonate with every risk manager who has filed a cyber claim, the Institute notes that in high-value facilities, financial losses from an interruption can exceed the physical damage itself. A melted transformer is a repair bill; the hours of dark racks behind it are the real loss.

The bottleneck isn’t capital — it’s confidence

Perhaps the most consequential line in the release concerns the supply side of the market. According to Swiss Re Institute, the principal constraint on insuring this boom is not the availability of insurance capital but the industry’s ability to deploy it confidently against exposures that are increasingly complex. Large infrastructure projects have limited operating histories, which makes loss frequency and severity genuinely hard to quantify; extreme accumulation potential complicates the diversification maths that reinsurance is built on.

Gianfranco Lot, Swiss Re’s Chief Underwriting Officer P&C Re, put it in terms the AI industry will recognise instantly: “We are seeing the digital economy become a real economy. AI needs data centres, power grids and increasingly complex infrastructure – and all of it needs insurance. That creates growth opportunities across multiple lines of business, but also significant risk concentrations. The deployment of capacity will depend on our ability to understand and manage those, and getting paid for the associated tail risk.”

Group Chief Economist Jérôme Haegeli framed insurance as a precondition for the build-out itself: “A new investment era is taking shape, with unprecedented amounts of capital flowing into the infrastructure that will power future economic growth. This also concentrates more value and creates new dependencies across power systems, supply chains and digital networks. Insurance is essential to making these investments resilient and financeable.”

The way forward

The report’s prescription is engineering-led underwriting, improved modelling and disciplined accumulation management, plus genuine global risk sharing — spreading exposures across insurers, reinsurers and capital markets so that no single balance sheet absorbs a USD 50 billion loss alone. Swiss Re points to precedent: the market has repeatedly adapted to emerging risks before, from nuclear power in the 1950s to cyber in the 2010s.

For the AI industry, the Monte Carlo message lands as a quiet reminder that the capex race has a financial plumbing layer underneath it. Hyperscalers can announce gigawatt campuses and trillion-dollar capex plans, but those projects remain financeable only if someone will insure them — and at a price that doesn’t break the model. The USD 200 billion premium pool is the insurance industry’s bet that it can learn to underwrite the AI era fast enough to collect it.

The English version of sigma 3/2026, Time to build: Expanding the frontier of insurability for the capex super-cycle, is available for download from Swiss Re’s website.