When finance wants to lock in 3-year savings and engineering knows the architecture is about to change
This is a hypothetical, illustrative scenario for teaching purposes, not an actual case study.
Company:
Manufacturing company with 5,000 employees, $800M annual revenue
Current State:
18 months into a 3-year cloud migration, $85K/month Azure spend
The Proposal:
CFO wants to buy 3-year Reserved Instances for VMs to save 60%
The Savings:
$612K/year if fully utilized (~$1.84M over 3 years—but only if the VMs run the full term)
The Concern:
Engineering plans to migrate to Kubernetes in 12-18 months
The Meeting:
Engineering needs to respond to the CFO's RI purchase proposal
The Context: Your company has successfully migrated 40% of applications to Azure and is running steady workloads on traditional VMs. The CFO's team analyzed the cloud spend and discovered that committing to 3-year Reserved Instances would save approximately 60% compared to pay-as-you-go pricing—that's over $600K per year in savings. However, your architecture team is actively planning a modernization initiative to containerize applications and move to Azure Kubernetes Service (AKS) within the next 18 months. If you commit to 3 years of VM-based Reserved Instances, you'll be locked into infrastructure you're planning to sunset.
Here's how this conversation usually unfolds when engineering takes a defensive posture
"I've reviewed our Azure spend. We're spending $85,000 per month on VMs. If we commit to 3-year Reserved Instances, we can save 60%—that's over $600,000 per year in savings. I'd like to move forward with this."
"That's not a good idea. We're planning to move to Kubernetes, so we won't need all those VMs."
"When is that happening?"
"Probably in the next 12 to 18 months. It's on the roadmap."
"So we'd have 18-24 months of savings. That's still worth over $900K. And honestly, IT projects always take longer than planned. Let's lock in the savings now."
"But if we commit to the wrong VM types or sizes, we'll be stuck with them. We need flexibility to optimize our architecture."
"Azure offers RI flexibility—you can exchange them or sell them. Plus we're under pressure to reduce IT costs. Our IT spend is 8% of revenue, and the board wants it under 6%."
"RI exchanges have restrictions. And selling unused RIs isn't as simple as it sounds. If we move to containers, we won't be able to use those VM reservations."
"I hear your concerns, but we need to show cost savings now. Unless you can give me a concrete plan with dates, I'm approving the RI purchase next week."
Result: CFO commits to 3-year RIs. 14 months later, the Kubernetes migration begins. The remaining 22 months of committed VM costs go unused, resulting in $748,000 of wasted spend. Engineering loses credibility.
Let's break down where the communication failed and what each party was really concerned about
Never presented the cost savings and business benefits of the Kubernetes migration as a competing investment
"12-18 months" and "on the roadmap" aren't concrete enough for finance to make decisions around
Never quantified the cost of being locked into the wrong infrastructure commitment
Could have suggested partial RI commitment for stable workloads while keeping flexibility for others
Didn't explain how committing now creates a sunk cost that will pressure delaying the modernization initiative
Now let's see how the same scenario plays out when engineering presents a financial comparison with concrete data
"I've reviewed our Azure spend. We're spending $85,000 per month on VMs. If we commit to 3-year Reserved Instances, we can save 60%—that's over $600,000 per year in savings. I'd like to move forward with this."
"I love the focus on cost optimization—$600K a year in potential savings is significant. I'd like to propose an alternative strategy that could save nearly as much with a fraction of the risk—let me walk you through the numbers."
"I'm listening. What's the alternative?"
"We have a funded Kubernetes modernization project scheduled to start in Q3 this year, with production cutover in Q2 next year—that's 12 months from now. Based on our analysis, containerizing our workloads will reduce our compute costs by approximately 55-65% compared to current VM-based infrastructure."
"Here's the financial comparison: 3-year RIs on our full VM footprint save $612K a year—about $1.84M over 3 years—but only if we keep those VMs running the whole time, and we won't. Once we migrate in 12 months, 24 months of that commitment is for capacity we no longer need—that's $816K locked into infrastructure we can't use. Net of that waste, the RI-only path really nets about $1.02M over 3 years. My proposal: 1-year RIs on the 40% of our footprint that isn't migrating—that's $204K in savings in year one, continuing every year after. Once Kubernetes goes live, the migrated 60% saves roughly $367K a year on top of that. Over 3 years, that's about $1.35 million in total savings, and none of it is wasted."
"But what if the Kubernetes project gets delayed? IT projects always slip."
"Good question. I've built a risk-adjusted scenario model. Even if the Kubernetes migration slips by a few months, this approach never locks us into wasted spend the way a 3-year RI commitment would—a delay just pushes out when the Kubernetes savings start. It doesn't strand any money we've already committed, which the full 3-year RI approach can't say."
"Additionally, to be clear about how we get there: we commit to 1-year RIs for our most stable workloads—about 40% of our current VM spend—and keep the rest on pay-as-you-go until cutover. This is the $204K in year-one savings I mentioned, growing to $571K a year once Kubernetes is live. It's less than the full 3-year commitment would nominally promise, but none of it gets stranded."
"What's the risk if we commit to 3 years and then the Kubernetes migration happens?"
"If we complete the migration in 12 months, we'll have 24 months of committed RI spend for infrastructure we're no longer using. That's approximately $816K of sunk costs we can't recover. Azure RI exchanges have limitations—they only apply to the same instance family, so we can't convert VM reservations to Kubernetes node capacity."
"There's also an opportunity cost: Once we've committed to 3-year RIs, there's organizational pressure to 'use what we paid for' even when a better option exists. This can delay strategic initiatives. I've seen this happen at my previous company—a $500K RI commitment delayed their cloud optimization by 18 months."
"You mentioned 55-65% savings from Kubernetes. How confident are you in that number?"
"Very confident. We ran a proof-of-concept with three of our production applications. We're seeing 60-70% reduction in compute costs for those workloads due to better resource utilization and auto-scaling. Industry benchmarks show similar results—Kubernetes typically delivers 50-70% infrastructure cost reduction. I'm using the conservative end of that range."
"This makes sense. Let's go with your hybrid approach: 1-year RIs for stable workloads, and funding for the Kubernetes migration on schedule. I want monthly progress reports on both the RI savings and the Kubernetes timeline."
"Perfect. I'll set up a cost tracking dashboard and provide monthly updates on both the committed savings and the migration milestones. If we hit any delays, we'll reassess the RI strategy at the 9-month mark."
Result: Hybrid approach approved. Immediate savings from 1-year RIs. Kubernetes migration funded and on schedule. 3-year total savings of $1.35M with zero wasted spend—versus the RI-only path's real net savings of about $1.02M after $816K in unavoidable waste. Engineering builds credibility with financial planning.
Here are the specific financial concepts the engineer used to transform the conversation
What it is: The cost of choosing one option over another
How it was used: "Committing fully to RIs means locking in $816K of waste instead of capturing $1.35M cleanly through the hybrid approach"
Why it mattered: Showed that the "safe" choice actually costs more money
What it is: Money already spent that can't be recovered
How it was used: "24 months of unused RI commitments = $816K sunk cost"
Why it mattered: Quantified the risk of locking into the wrong infrastructure
What it is: The point at which benefits equal costs
How it was used: "Even with a delayed migration, the hybrid approach never strands committed spend the way a 3-year RI does"
Why it mattered: Showed the margin for error and reduced perceived risk
What it is: Comparing multiple possible futures with different assumptions
How it was used: "3-year RI: ~$1.02M real savings after $816K in waste vs. Hybrid + Kubernetes: $1.35M savings with none wasted"
Why it mattered: Gave finance a clear financial comparison between options
What it is: Comparing returns of competing investment options
How it was used: "Kubernetes delivers better ROI than RIs over 3 years"
Why it mattered: Made the modernization project financially attractive
What it is: Factoring in probability of delays or failures
How it was used: "Even with a delay, the hybrid approach never locks in wasted spend the way the RI commitment would"
Why it mattered: Addressed finance's skepticism about IT project timelines
Instead of saying "no" to RIs, proposed a hybrid approach with $1.35M in savings and zero wasted spend, vs. the 3-year RI commitment's ~$816K of guaranteed waste. This reframed the conversation from "resist spending" to "invest smarter."
"Q3 start, Q2 next year cutover" is specific enough for finance to plan around. "12-18 months maybe" is not.
$816K of wasted RI spend made the risk concrete. This transformed "we might need flexibility" into "we could waste $816K."
Built risk-adjusted scenarios showing that even with 6-month delays, the approach still beats pure RIs. This directly addressed finance's skepticism.
1-year RIs for stable workloads gave finance immediate wins while preserving flexibility. Shows business awareness and willingness to meet in the middle.
60-70% cost reduction from POC testing plus industry benchmarks made the Kubernetes savings credible, not speculative.
Analyze your own infrastructure and evaluate the RI vs. modernization trade-off:
What's your current monthly cloud spend on compute (VMs, instances)? Calculate both the annual and 3-year savings from committing to 3-year Reserved Instances at 60% discount—make sure you're clear on which one you're citing.
Do you have any planned architectural changes in the next 24 months (containers, serverless, multi-cloud, etc.)? If so, when and what percentage of workloads would be affected?
If you modernize your architecture, what cost savings do you project? (Use industry benchmarks: Kubernetes = 50-70%, Serverless = 40-60% for applicable workloads)
Compare total 3-year savings: (A) 3-year RIs, (B) 1-year RIs + architectural modernization, (C) No commitment + immediate modernization. Which scenario saves the most money?
Calculate the sunk cost risk: If you commit to 3-year RIs and then migrate in 12 months, how much committed spend would go unused?
Bonus Challenge: Create a decision matrix spreadsheet with columns for: (1) Scenario name, (2) Year 1 cost, (3) Year 2 cost, (4) Year 3 cost, (5) Total 3-year cost, (6) Sunk cost risk, (7) Flexibility rating. Use this to present options to your CFO.
You've completed all the finance topics. Keep referring back as needed.