In 2026, the lines between Site Reliability Engineering (SRE) and infrastructure teams are blurring as infrastructure companies enter the observability space. They’re not just offering better dashboards; they’re aiming for a comprehensive control over telemetry data, which allows them to package additional services. Recent acquisitions, like Palo Alto Networks buying Chronosphere and Snowflake acquiring Observe, highlight this shift. These moves indicate a strategic pivot from merely selling tools to engaging in a broader entry-point competition within the observability market. Observability is expensive and complex, with real costs associated with incidents and slow recovery times. Many teams are frustrated with unpredictable pricing, and companies are actively seeking better cost management solutions. This frustration has led to job postings explicitly calling out the need for control over observability costs. However, the high costs alone don’t explain why platform companies are making major acquisitions in this area. The integration of AI plays a significant role; if AI can reduce mean time to recovery (MTTR) and automate some incident response tasks, teams see it as a worthwhile investment. The article emphasizes the role of OpenTelemetry as a standard for observability. It decouples the instrumentation from the backend, easing transitions between systems. The author suggests that the future standard stack will combine OpenTelemetry with Apache Iceberg, which provides an open data layer, reducing vendor lock-in. This flexibility is crucial, especially in AI scenarios where multiple agents might need to run parallel queries. The potential for skyrocketing costs is a real concern when using vendor-specific systems, especially if agents generate inefficient or erroneous queries that could strain resources during critical incidents. Open table formats are becoming essential for managing these challenges effectively.