We Spent $2 Million Leaving AWS. Then We Spent $3 Million Coming Back.

2026-03-06 · Source: Machine Learning on Medium · Field: Technology & Digital — Cloud Computing & IT Infrastructure, Cloud Economics & Strategy · Depth: Intermediate, quick

Summary

A company spent $2.1 million over 18 months to repatriate its entire infrastructure from AWS to a colocation facility, driven by the belief that cloud services were too expensive and on-premise solutions would yield significant savings. This involved purchasing physical servers, negotiating bandwidth, and hiring a dedicated operations team. However, by February 2026, the company reversed course, returning to AWS at an additional cost of $3 million, illustrating what they term the "Cloud Exit Paradox." Initial spreadsheet projections had suggested a 60% annual savings by moving off AWS, with an annual AWS bill of $1.4 million compared to $600K for hardware amortization and $200K for colocation space and power.

Key takeaway

For Directors of Infrastructure or VPs of Engineering considering cloud repatriation to reduce costs, recognize that the "Cloud Exit Paradox" can result in far greater expenses than anticipated. Your team should conduct a comprehensive, multi-year total cost of ownership analysis that includes migration, operational overhead, and potential re-migration costs before committing to an exit strategy.

Key insights

Repatriating cloud infrastructure can lead to significant, unforeseen costs, often exceeding initial cloud expenses.

Principles

• Cloud repatriation carries substantial hidden costs.
• Initial cost projections often underestimate complexity.

In practice

• Thoroughly audit all hidden costs of infrastructure migration.
• Evaluate total cost of ownership beyond direct hardware/colo.

Topics

• Cloud Repatriation
• AWS Costs
• Cloud Exit Paradox
• Infrastructure Migration
• On-Premise Infrastructure

Best for: VP of Engineering/Data, Director of AI/ML, Entrepreneur, CTO, Executive, IT Professional

Related on AIssential

• What Your LLM Integration Actually Costs Per Token — The true cost of LLM integration is 3-10x the API price due to an unbilled "tax stack" of operational overhead.
• Why cloud value depends on what happens after migration — Cloud value stems from post-migration strategy, governance, and optimization, not just workload movement.
• I Found 487 GB Of Data Nobody Had Looked At In Three Years. Deleting It Felt Like Archaeology. — Regular storage audits can identify dormant data, reducing costs and improving database efficiency.
• Our SaaS Contract Renewal Came In 18% Higher. We Paid It. Then We Built the Exit Plan. — Proactive planning is crucial to avoid vendor lock-in and manage escalating SaaS costs.
• Tech industry averages just 5% GPU utilization, report finds — Widespread GPU and CPU overprovisioning leads to significant financial waste in tech infrastructure.
• Measuring the Hidden Costs of AI-Generated Insights: A Data Analyst’s Guide to Autonomous Pipeline… — Autonomous AI pipelines accrue hidden, unbudgeted costs over time, impacting long-term ROI.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning on Medium.