Introducing 18-Month Forecasting and Explainable AI Insights in AWS Cost Explorer

We’re excited to announce enhanced forecasting capabilities in AWS Cost Explorer, now providing up to 18 months of future cost projections with improved accuracy and AI-powered explanations (preview). This extended forecast horizon addresses the need for long-term financial planning that aligns with enterprise fiscal cycles, while providing transparency into the drivers behind your cost forecasts.

The challenges of enterprise cloud financial planning

AWS Cost Explorer helps you understand your historical cloud spending patterns and provides forecasts for future spend, enabling you to analyze costs across services, accounts, and time periods. Previously, AWS Cost Explorer has provided up to 12 months of future forecasting capabilities, using machine learning models that analyze six months of historical data to generate projections for your cloud spending patterns. This foundation has served many organizations well for quarterly planning cycles and short-term budget allocation decisions, enabling FinOps teams to anticipate costs and optimize spending.

However, as cloud adoption matures and becomes central to business operations, organizations face key challenges that extend beyond these current capabilities. Enterprise financial planning operates on extended cycles that often begin in fiscal quarters Q3 and Q4, requiring cost visibility that extends 15-18 months into the future. While the current 12-month window works well for many planning scenarios, annual planning seasons benefit from extended visibility beyond 12 months to cover the final quarters of the fiscal planning cycle. Additionally, with current forecasting models analyzing six months of historical data, models miss long-term patterns that impact cloud spending. Seasonal business cycles, holiday traffic spikes, fiscal year-end processing loads, and annual compliance activities create spending patterns that repeat yearly but can be hard to capture with a shorter historical analysis window. This limited historical context can lead to forecasts that don’t account for your organization’s unique cyclical spending behaviors. Finally, when presenting cloud forecasts to executive leadership, your FinOps teams need to explain what factors are driving projected costs and justify their predictions to help executives make informed decisions about cloud investments and optimization opportunities. Current forecasting provides the “what” but not the “why” behind cost projections, making it difficult to build confidence in forecast projections or identify opportunities for optimization based on the underlying drivers of cost growth.

Extended forecasting with enhanced intelligence

AWS Cost Explorer now addresses these challenges with expanded forecasting capabilities that align with enterprise planning cycles. The new 18-month forecasting horizon provides the extended visibility that you need for annual budget planning. This extended timeline provides the complete visibility needed for your longest-range planning scenarios, ensuring data-driven confidence across your fiscal planning cycle.

The enhanced machine learning model can now analyze up to 38 months of historical data once you opt-in, a six-fold increase from the previous six-month window limit. The model weighs this historical data, giving greater emphasis to recent months while using older data to identify longer-term patterns and seasonal trends. This expanded historical context enables the model to identify and learn from your organization’s unique spending patterns, including seasonal variations, annual compliance cycles, holiday traffic patterns, and fiscal year-end activities. By understanding these longer-term cycles alongside recent trends, the forecasting engine can make more informed predictions about future spending that account for both the cyclical nature of your business operations and recent architectural or usage changes. For example, if your organization consistently experiences increased compute costs during holiday seasons or end-of-quarter processing periods, the updated forecast model will anticipate and account for these predictable variations while still reflecting recent shifts in your cloud infrastructure or business model.

Understanding your forecasts with AI-powered explanations (launching as preview)

Beyond extended forecasting horizons and enhanced accuracy, AWS Cost Explorer now provides GenAI-powered natural language explanations that detail the drivers behind your forecast predictions. This new AI capability adds transparency to cost forecasting, delivering clear, understandable insights that FinOps teams can use to engage stakeholders and use for strategic decision-making.

When you view your AWS forecast, you will receive a clear, natural language explanation summary that breaks down the primary factors influencing your predicted costs. These explanations might highlight seasonal patterns the model has identified, such as increased storage costs during quarterly data archiving periods, or compute spikes that correlate with your organization’s annual processing cycles. The AI explanations can also surface unexpected trends, such as gradual shifts in service usage patterns or the impact of recent architectural changes on long-term cost projections.

The explainable AI feature is available as a preview, allowing AWS to gather user feedback and continuously improve the quality and relevance of the insights provided. These AI-powered explanations are available exclusively through the AWS Cost Explorer console, providing an interface for exploring the reasoning behind your forecasts. Whether you’re preparing annual budget presentations, or investigating unexpected forecast changes, the natural language explanations help bridge the gap between machine learning predictions and practical business insights that drive financial decisions.

Getting started

The extended 18-month forecasting capabilities are available now in AWS Cost Explorer at no additional cost, accessible through both the Cost Explorer console and APIs. The AI-powered explanations are now available in preview, accessible exclusively through the AWS Cost Explorer console to gather user feedback and continuously improve the insights provided.

To get started with the new capabilities, follow these steps:

Step 1: Open Billing & Cost Management → AWS Cost Explorer. As shown in Figure 1, a flash bar notification appears at the top of Cost Explorer to inform customers about the new forecast explainability capability.

Figure 1. Flash bar notification in AWS Cost Explorer console announcing the new AI-powered forecast explainability feature

Step 2: Choose a report (e.g., Monthly cost by Service) → Report Parameters → Forecast Horizon. Future forecasts can now extend up to 18 months into the future, as shown in Figure 2, providing the long-term visibility enterprise teams need.

Figure 2. AWS Cost Explorer Report Parameters showing the extended 18-month forecast horizon option in the menu

Step 3: To review narrative insights, cost contributors, and seasonality patterns, click the “Generate forecast explanation” button (Figure 3) that appears when you enable a forecast in the console.

Figure 3. Generate forecast explanation’ button displayed in the AWS Cost Explorer interface after enabling forecast a forecast

Step 4: Understanding Your Forecast. AI-powered natural language explanations detail the key drivers behind your forecast, as shown in Figure 4.. We encourage you to explore the explanations and provide feedback through the console interface to help us refine and improve the insights over time.

Figure 4. AI-powered forecast explanation panel displaying natural language insights about cost drivers, contributors, and seasonality patterns with feedback options

Prerequisite: For optimal forecasting accuracy, we recommend enabling 38-month data retention in Cost Explorer. This extended retention period provides the machine learning model with the full historical context needed to identify seasonal patterns, cyclical trends, and long-term spending behaviors that significantly improve forecast precision. To enable 38-month retention, visit the Cost Explorer preferences page in your AWS Billing and Cost Management console and adjust your data retention settings. If you choose not to enable 38-month retention or don’t currently have extended historical data available, the forecasting model will use whatever data is accessible, defaulting to the standard six-month historical window. While you’ll still benefit from the 18-month forecasting horizon and AI explanations, the enhanced pattern recognition capabilities work best with the full 38-month historical dataset.

Conclusion

Enable these enhanced forecasting capabilities today to experience the improved accuracy and extended visibility that comes from improved pattern recognition and explainable AI insights. Learn more about AWS Cost Explorer forecasting capabilities in our documentation and start planning your cloud investments with the extended horizon your organization needs.