Post by Alexey Vidanov

re:Invent has just begun, and AWS came out strong with a strategic announcement: AWS Lambda Managed Instances ⚡ This update changes how companies think about compute costs, architecture choices, and cloud operating models. What problem does it solve? 🎯 Classic Lambda excels at unpredictable workloads. But once executions stretch toward minutes (max 15 min per invocation), costs grow, and teams switch to EC2 or containers. Managed Instances remove this dilemma: - Lambda experience + EC2 control - You keep event-driven development and automation, but functions run on dedicated EC2 capacity you own. Pricing made simple 💰 You now pay for: 🟢 $0.20 per 1M requests 🟢 EC2 instance cost you choose (On-Demand, Savings Plans, RIs apply) 🟢 +15% management fee based on the EC2 On-Demand rate 🔵 No duration charges This transforms Lambda from a “millisecond billing engine” into a predictable EC2-based model. For long-running or data-heavy workloads, this can be a substantial reduction in monthly spend. Why you should care 🧭 ✔️ predictable budgets ✔️ one operational model instead of Lambda + ECS clusters ✔️ reuse of existing EC2 commitments and discounts ✔️ suitable for ML inference, streaming, ETL, and other sustained workloads ✔️ keeps the familiar 15-minute execution boundary, but removes execution-time billing pressure This is compute alignment, not just another feature drop. Strategic takeaway 🚀 AWS is closing the gap between serverless convenience and infrastructure economics. Lambda Managed Instances bring stable, long-running workloads into the serverless world without unpredictable cost curves. If this is the opening move of re:Invent, the week will be interesting.