The Challenge
The cyber insurance market is at a pivotal moment. GlobalData projects it will reach $35.4 billion by 2030, up from $22.2 billion in 2025. This 59% increase in five years presents a challenge: traditional risk models can't keep up with the volume and complexity of new cyber exposures.
Climate-related losses are also rising, making historical loss data unreliable. Severe weather events are rendering traditional actuarial tables obsolete in some regions, making them uninsurable under conventional models.
The issue isn't just growth; it's that old tools are ineffective. You can't price ransomware risks with outdated data when attack vectors change monthly. You can't model wildfire exposure using 30-year histories when fire seasons now last twice as long.
Pressures on Underwriters
Underwriters face three main pressures:
Volume constraints: The expanding cyber insurance market means you're evaluating more applications and complex technology stacks than your team can manually assess.
Data constraints: Climate change has disrupted the assumption that past loss patterns predict future risks. For cyber, you're pricing risks that didn't exist 18 months ago.
Competitive constraints: The industry is consolidating around AI capabilities. In 2025, M&A deals within AI in insurance grew 328% in value and 125% in volume. Your competitors are acquiring capabilities you may need to develop.
This creates a bind for actuaries: your standards require credible data and sound methods, but the available data is increasingly unreliable for future risk assessment. Climate models project conditions outside historical ranges, and cyber threat intelligence describes attack patterns with no loss history.
Munich Re's Strategic Approach
Munich Re's July 2025 acquisition of Next Insurance shows one response to these constraints. Instead of adding AI to existing processes, they acquired a company built around AI and digitalization.
This decision highlights a strategic choice: treat AI as infrastructure, not just a feature. Next Insurance didn't just use machine learning for claims; they built their underwriting around AI's ability to process live data and adapt to changing conditions.
The industry's shift toward agentic AI reflects similar logic. These systems don't just analyze static data; they react to live information and adjust decisions as conditions change. For underwriters, this means:
Dynamic risk scoring: Agentic systems can update risk scores as your insured's security posture changes or as new vulnerabilities emerge.
Real-time climate modeling: AI can use current satellite data and weather patterns to price natural catastrophe exposure based on forward-looking conditions.
Automated control verification: For cyber risks, agentic AI can continuously verify that required security controls are in place, flagging lapses that alter your risk profile mid-term.
Results and Metrics
The industry's investment in AI capabilities led to measurable shifts in 2025. The 328% increase in M&A deal value within AI in insurance wasn't speculative; it showed carriers recognizing that their existing infrastructure couldn't support the growth and complexity they faced.
The cyber insurance market's projected growth hinges on solving the underwriting scalability problem. You can't manually underwrite $35 billion in premiums using processes that supported a $22 billion market.
For natural catastrophe risk, the results are evident in market withdrawals. Carriers are exiting regions where traditional models can't produce profitable rates. The AI investment thesis is that better modeling can identify profitable segments or price risk accurately enough to remain in markets conventional approaches would abandon.
Addressing Gaps
The industry's current approach has a gap: it's focused on deployment speed without enough attention to model governance. Agentic AI systems need real-time oversight frameworks, which most carriers lack.
If you're building AI-driven underwriting now, invest in:
Explainability infrastructure: Actuaries need to understand why a model priced a risk a certain way. For cyber risks, you'll face regulatory questions about algorithmic decisions.
Validation protocols: Develop proxy validation methods and continuous monitoring frameworks for models projecting future conditions.
Human override procedures: Define when underwriters should intervene in AI decisions. Set these boundaries before deployment.
Munich Re's approach of acquiring AI capability sidesteps some technical risks but creates integration challenges. If you're considering a similar move, plan for a two-year integration timeline. Merging underwriting cultures is harder than merging codebases.
Takeaways for Your Team
Rethink your data strategy: If you're still using historical loss data, you're pricing yesterday's risks. Invest in forward-looking data sources: threat intelligence feeds for cyber, climate projection models for property, and real-time security posture data for technology risks.
Separate AI hype from capability: The 328% increase in deal value reflects genuine capability gaps, but not every AI vendor solves your problem. Test whether a tool improves your loss ratio before scaling it.
Build governance before automation: Agentic AI will make thousands of decisions your team used to make manually. Prepare audit trails, override protocols, and explainability frameworks before deploying autonomous systems.
Price for non-stationarity: Both cyber and climate risks defy the assumption that future losses will resemble past ones. Your pricing models need to incorporate scenario analysis and forward-looking risk factors.
The convergence of AI capability and accelerating risk complexity is the current operating environment. Your competitors are already building capacity you may still be planning. The question isn't whether to adapt your underwriting infrastructure. It's whether you're moving fast enough.
