Insurers must look beyond AI growth to understand risks and build expertise: VIG Re

The AI infrastructure boom is expected to bring huge new values into the insurance market, but VIG Re’s Aleš Mirnik argues that understanding those risks may prove as difficult as finding the capacity to cover them.

Key points:
Complexity needs deeper expertise
Connected risks raise accumulation
Data still needs judgement

The re/insurance industry is excited about the huge investment flowing into data centres, seeing them as a major growth opportunity. But as these facilities become more complex, there is a danger that the appetite to write the business runs ahead of the industry’s ability to understand the risk.

As reinsurers gather in Baden-Baden with AI and data centre growth high on the agenda, Aleš Mirnik, head of CEE & East Mediterranean at VIG Re, is urging the market to look beyond premium and capacity and spend more time understanding what sits behind the exposure.

“Quite a lot of the time, industry professionals see the premium, the volume and the growth. They don’t see the complexity,” he told Baden-Baden Today.

Part of that complexity comes from the speed at which the technology itself is changing. Semiconductor production is moving through successive generations of smaller and more sensitive technology, while the equipment used to manufacture chips can be vulnerable to relatively minor disruption.

“Technology is advancing faster than the infrastructure built around it,” Mirnik said. “The complexity is increasing literally day by day.”

He pointed to cases where an earthquake caused little or no direct damage to semiconductor machinery, but triggered sprinklers that damaged sensitive production equipment.

“You almost need to be a scientist to assess whether a production line will face a delay, a complete shutdown or damage, and whether you need to replace everything or just one part,” he said. “This is now a completely different game.”

“As an industry, we will need more specialists working alongside insurance companies to assess these risks,” Mirnik said.

Scale of risk

Technical complexity is only one part of the problem. The values concentrated at individual locations are already stretching available insurance capacity.

Some of the largest semiconductor facilities and data centres carry values running into tens of billions of dollars, while available cover may represent only a fraction of the exposure.

“It will take the whole market to insure this technology space, and that is only the property exposure,” Mirnik said.

“Then there is business interruption and contingent business interruption, because a whole chain of industries may depend on a particular data centre.”

Data centres also rely on power, cooling, water and telecommunications, while many customers can depend on the same facility. Some planned AI campuses are approaching gigawatt scale.

“The underlying perils may be familiar, but their scale and interdependence take the analysis to another level,” Mirnik said.

That means looking beyond buildings and equipment to water availability, fire protection, power and cooling redundancy, cyber security, contractual obligations and accumulation across portfolios.

“Commercial opportunities are substantial, but sustainable capacity in such a fast-changing environment requires specialist expertise, regular reassessment and a multidisciplinary understanding of the risk,” he said.

“Commercial opportunities are substantial, but sustainable capacity requires specialist expertise.”

Risks connect

The same concern runs through Mirnik’s view of emerging risks more broadly. Technology, climate change and geopolitics will all shape insurers’ portfolios, but the greater challenge is how those risks interact.

“A geopolitical conflict can disrupt energy supplies and trigger cyber events; extreme weather can affect infrastructure and supply chains; and a technology failure can interrupt businesses across several countries at once,” he said.

For a reinsurer, the question is not only how severe an individual loss could be, but how far it could spread across policies, portfolios and lines of business.

“The greatest emerging risk may not be one specific peril but failing to recognise how different risks connect and accumulate,” Mirnik said.

Climate risk adds to that challenge. Changing weather patterns, asset values and concentrations mean model output must be tested against recent experience, alternative scenarios and expert judgement.

“Models structure our understanding of risk; technical underwriting determines how we act on it,” he said.

Underwriting judgement

AI itself can help insurers handle some of that complexity. VIG Re sees value in using it to organise information, identify patterns and reduce routine work, giving underwriters more time for complex risks and portfolio management.

But Mirnik draws a clear line around responsibility. “AI should strengthen technical underwriting without weakening accountability,” he said. “The underwriter must understand, challenge and ultimately own the decision.”

He emphasised that good underwriting is “not a choice between data and judgement” but comes from combining the two. Imperfect data does not automatically rule out a risk, but uncertainty must be reflected in assumptions, pricing or structure.

As investment in AI infrastructure grows, the harder task, Mirnik said, will be understanding those risks well enough to know where capacity should be deployed and how far a loss could spread beyond the original site.

Aleš Mirnik is head of CEE & East Mediterranean at VIG Re, and can be contacted at: a.mirnik@vig-re.com.

For more news from Baden-Baden Today, click here.

Did you get value from this story? Sign up to our free daily newsletters and get stories like this sent straight to your inbox.