人工智能在保险及再保险领域的风险与机遇 Q2：数字基础设施

Artificial Intelligence — Risks and Opportunitiesin (Re)insurance and BeyondQ2 — Digital Infrastructure This report is a collaborationbetween Gallagher Re,Gallagher and CB Insights. Contents I N S I D ET H I S E D I T I O N . . . 04.IntroductionQ2 InsurTechInvestment Data Highlightsand Foreword•Dr. Andrew Johnston, Gallagher Re Case Studies24. 16. •Parametrix•Advanced Technology Assurance Ltd. Deal ofthe Quarter•Shepherd34. 26. 38. Insuring the AI Backbone:How the Data CenterBoom Is Reshaping the(Re)insurance Landscape •Lara Mowery and Luca Drane,Gallagher Re 54.The Data Center•The quarter’s data highlights 50. 46. Incumbent Corner•Chris Dempsey, FM Intellium •Kyle Nakatsuji, Dearborn Labs Introduction Q2 InsurTech investmentdata highlights and foreword DR. ANDREW JOHNSTON In 2026, we are examining the insurance risks and opportunities that the rise ofAI is creating. In Q1 we assessed the topic of AI liability risk and its overlap withcyber insurance, among other risks emanating from digital exposures. In thisQ2 edition, we turn our attention to AI infrastructure — the machines, storageand piping required to keep pace with the growing demand for sophisticatedcomputing power. Specifically, much of our focus will be on data centers. two and a half years. The median InsurTech deal size reachedUSD10.0M in Q1’26, almost double the USD5.3M seen at the heightof the venture-funding boom in 2021. Yet, at the same time,InsurTech deal count fell to just 81 in Q1’26 — the lowest since Q2’16(67 deals). Capital availability is clearly not a problem; InsurTechfunding is rebounding. However, if the money is going to a smallernumber of innovators, does this leave incumbent carriers withfewer options? Before proceeding to the report’s main theme, however, we firstwant to note some important developments in the InsurTechfunding environment. There is a risk that despite the generalrecovery in fundraising we have witnessed in recent months,the industry may be facing a narrowing of its innovation pipeline. The final quarter of 2025 and the first of 2026 witnessed a materialjump in funding, to an average USD1.65 billion a quarter — froma long-run average of USD1.1 billion a quarter over the preceding Introduction There is also a human element to this story. In the lead-up tothe funding peak of 2021, it was not unusual to see talent movefrom established insurance businesses to InsurTech start-ups.Consequently many InsurTechs were well-staffed, with accessto experienced (re)insurance professionals. While it would beuntrue to say that we never hear of such things happening today,it is far less frequent. In fact, we often observe InsurTechemployees going in the opposite direction. To some extent thisis a sign of a mature industry, and unsurprising — but it could alsobe a limiting factor on any proliferation of new start-up ideas. Is AI taking the (Insur)Tech jobs? More broadly in society, there is a (not irrational) fear of AIreplacing certain jobs — or even leading to mass unemployment.If we take history as a guide, however, the AI boom shouldultimately be job-creating. In the industrial revolution, factoriesreplaced farms and machines replaced manual labor, leadingto job losses and social unrest. But in the long term, significantlymore jobs were created than lost, in a range of new industriesfrom textiles to mining and iron and steel production. In morerecent times, the internet revolution also disrupted many industriesand led to the closure of brick-and-mortar businesses. Butoverall, much like the industrial revolution, it was job creating.It is estimated in the last 20 years, companies like Google, Amazonand Facebook have added three million jobs. The e-commerce andassociated logistics industry has boomed, and the so-called ‘gigeconomy’ has been created. Some studies suggest that, in certainsectors, the internet has added 2.5 jobs for every job lost.We expect that AI will be no different. While the game ischanging, the players and the prizes are not. This leads us to consider what start-ups are spending their moneyon — if not people. Historically, InsurTech company valuationsand funding targets were primarily based upon the perceivedtotal addressable market (which was often very ambitious)coupled with the current size of an organization (its headcount),and its projected ability to continue to hire in and assimilate newtalent. However, in the AI-first era, it is many people’s assumptionthat tech companies will require fewer staff.1 Offsetting that, perhaps, is the reality that AI tools are generallymore expensive to build than non-AI tools, and therefore thereis a greater requirement for capital. The difference depends heavilyon the complexity, data requirements and integration depth ofthe project. AI generally requires specialized skills (think machinelearning engineers), rigorous data cleaning and preparation,and in some cases, very powerful new infrastructure. However,for the innovation to make sen