Beyond backups: Six lessons on cloud resilience, recovery, and rising AI costs

In a recent roundtable discussion at TechSparks 2026, cloud-native companies discussed the evolving challenges of cloud resilience, recovery, and AI costs. Sakthivel Murugan, Head of IT at DevOn, and Althaf Kudiyama Kandi, Director of Engineering at Moody’s Analytics, were among the speakers. The experts emphasized that ensuring quick restoration of critical services has become increasingly important as AI, automation, and complex infrastructure add new risks.

Resilience is no longer just an internal technology concern; it influences customer trust and platform selection. Companies moving from perpetual software licences to SaaS platforms have higher expectations around security, governance, and availability. As a result, businesses are rethinking their approach to reliability, as recurring monthly downtime has led one company to prioritize it, achieving 99.5% availability and targeting 99.9% by March 2027.

Backups are only as good as recovery drills, and organizations should conduct frequent and comprehensive testing. Coordinating exercises across various systems can be challenging, and testing should focus on restoring entire businesses, not just individual components. Cyber recovery needs its own drills since cyberattacks are frequent, but testing often does not receive the same attention as disaster recovery.

Another key takeaway is that rising cloud costs are becoming more difficult to manage due to AI workloads and currency fluctuations. Companies should trace costs to individual services, make cloud spending a key performance indicator for teams, and balance cost-saving measures with reliability. AI brings new demands to security and infrastructure, and businesses must address these without allowing costs to grow unchecked.

However, reducing infrastructure spending too aggressively can lead to larger risks if critical systems fail.

Finally, recovery plans must align with a business's priorities. Essential service providers must prioritize continuity, while others are moving towards active-active configurations for critical services. The four pillars of resilience—prevention, detection, response, and recovery—must work together to provide effective protection.