1-Year vs 3-Year Savings Plans: The Decision Framework
Three-year Savings Plans deliver 30–45% deeper discount than one-year plans. The math favors three years for committed baseline capacity, but the optionality cost is real. The buyer-side framework for choosing.
The 1-year vs 3-year choice is one of the most consequential decisions in a Savings Plans purchase. The discount tier gap is large — typically 30 to 45 percentage points of additional savings against On-Demand. On a $3M Savings Plan, that's roughly $900K to $1.35M of additional savings over the term, all else equal.
All else is not equal. The 3-year commitment locks the buyer into a fixed dollar-per-hour spend for three years; the 1-year commitment runs out in twelve months. Across 500+ engagements and $2.4B+ in reviewed AWS spend, the right choice is usually 3-year for the long-term baseline layer and 1-year for everything else — but the analysis behind that prescription is more nuanced than the headline.
The discount gap
The published discount tiers for Compute Savings Plans:
| Term & Upfront | Discount vs On-Demand | Delta vs 1-yr No Up |
|---|---|---|
| 1-yr, No Upfront | ~17% | baseline |
| 1-yr, All Upfront | ~24% | +7 pts |
| 3-yr, No Upfront | ~50% | +33 pts |
| 3-yr, All Upfront | ~58% | +41 pts |
EC2 Instance Savings Plans show even larger gaps — the 3-year All Upfront tier can reach 72% off On-Demand, vs ~35% for 1-year All Upfront.
Read purely on discount, 3-year wins decisively. The interesting question is what the 1-year option buys you in exchange for the discount sacrifice.
What 1-year commitment buys
The 1-year option preserves three forms of optionality:
1. Cloud-strategy optionality
Buyers running active multi-cloud evaluations want the option to shift workload composition between AWS, Azure, and GCP. A 3-year AWS commitment locks usage to AWS regardless of how the multi-cloud evaluation evolves. A 1-year commitment expires before most multi-cloud transitions complete.
2. Architectural optionality
Major architectural shifts (Graviton migration, EKS adoption, serverless transition, region consolidation) change workload composition. A 3-year commitment locks current architecture into the commitment baseline; a 1-year commitment expires in time for renewal to reflect the new architecture.
3. Business-model optionality
Startups in active product-market fit search, enterprises in active restructuring, businesses with M&A in flight — these have higher probability of fundamental business-model change within three years. The 1-year commitment carries forward less commitment risk.
The commitment risk math
The right framework for evaluating the 1-year vs 3-year choice is risk-adjusted expected value. The 3-year commitment delivers more discount; it also carries more commitment risk. The question is whether the discount delta exceeds the expected cost of commitment risk.
Expected cost of commitment risk = probability of usage shortfall × magnitude of unused commitment × commitment value
For a typical workload baseline:
- Probability of usage shortfall (the commitment being larger than realized usage): 5–15% for stable enterprise workloads, 15–30% for growth-stage or rapidly-changing workloads.
- Magnitude of unused commitment when shortfall occurs: typically 10–20% of committed value.
- Commitment value: full discounted commitment dollars across the term.
For a stable enterprise workload with 10% shortfall probability and 15% shortfall magnitude, expected cost of commitment risk is roughly 1.5% of committed value. Compared to a 33-point discount gap, the 3-year option wins decisively.
For a growth-stage workload with 25% shortfall probability and 20% shortfall magnitude, expected cost of commitment risk is roughly 5% of committed value. The math still favors 3-year, but the margin is narrower, and the structure of the decision becomes more important than the binary choice.
Across the 500+ engagements we review, the median buyer commits ~60% of total Savings Plans to 3-year and ~40% to 1-year. Buyers who default-to-1-year capture roughly 18–24% blended discount; buyers who layer 3-year on baseline and 1-year on the flexibility tier capture roughly 31–38% blended discount. The gap is the cost of defaulting to 1-year on the irreducible baseline.
The layering approach
Rather than choosing 1-year or 3-year for the entire commitment, the right structure is layered:
- Long-term baseline layer (3-year): Sized to the growth-down 25th percentile hour — the part of the workload that will be present in three years under nearly all forecast scenarios.
- Medium-term layer (1-year): Sized to absorb usage between growth-down baseline and base-case expected average.
- Elastic remainder: On-Demand or Spot.
The 3-year layer captures the deep discount on the part of the baseline that genuinely is committable for three years. The 1-year layer captures the discount on the medium-term portion without locking in long-duration commitment risk.
The three-year confidence test
For each 3-year layer purchase, run the confidence test:
- Project the workload three years forward at the growth-down scenario.
- Identify the floor of usage that the workload will support under that scenario.
- The 3-year commitment level should be at or below that floor.
If the buyer cannot confidently project a three-year usage floor — because cloud strategy is unsettled, architecture is in flight, or business model is uncertain — the 3-year layer should be smaller than typical. The 1-year layer absorbs more of the baseline; total discount is lower, but commitment risk is appropriately managed.
Workload profiles where 3-year layer should be smaller
- Active multi-cloud evaluation with decision-point inside three years.
- Active Graviton migration that will shift ~30%+ of compute to ARM-based instances inside three years.
- Active major-region consolidation that may change region mix substantially.
- Major M&A integration in flight.
- Restructuring or divestiture under consideration.
Workload profiles where 3-year layer can be larger
- Mature production workloads with low rate of architectural change.
- Inference workloads supporting stable production traffic patterns.
- Data-intensive workloads with high data-gravity costs of migration off AWS.
- Regulated workloads with compliance certifications that would be expensive to re-validate elsewhere.
The renewal timing consideration
1-year and 3-year Savings Plans have different renewal cadences, which affects portfolio operations:
- 3-year plans: Renew every three years. Lower operational overhead, but each renewal is a large decision.
- 1-year plans: Renew annually. Higher operational overhead, but each renewal is a smaller decision and can adjust to current trajectory.
Mature buyers stagger renewal dates across the portfolio — not all plans expiring on the same day. Staggered renewals smooth the operational cadence and reduce the size of any single renewal cliff.
This is particularly important for the 3-year layer. A buyer with $5M of 3-year Savings Plans all expiring on the same day faces a $5M renewal decision in a single quarter. Staggering across four or five purchase dates spreads the renewal load and provides more data points for sizing decisions.
The EDP interaction
Savings Plans commitments count toward EDP commit at the same rate regardless of term. A $3M annual 3-year Savings Plan and a $3M 1-year Savings Plan contribute the same $3M annual EDP burn for the years they overlap with the EDP term.
This means the term choice is genuinely independent of EDP positioning. Run the discount-vs-optionality calculation on its own terms.
One subtle interaction: if an EDP renewal is scheduled inside the 3-year Savings Plans window, the Savings Plans commitment becomes part of the negotiating context for the EDP. AWS will know you have committed dollar-per-hour spending across the EDP term, which affects the discount tier negotiation. Mature buyers structure Savings Plans purchases to align with EDP renewal cadence — typically purchasing major 3-year Savings Plans either well before or well after EDP renewal events.
The interest rate consideration
3-year All Upfront commitments deploy capital for three years; 1-year All Upfront commitments deploy capital for one year. In rising-rate environments, future capital becomes more valuable, and longer-duration capital deployments become commercially worse. In falling-rate environments, the opposite.
For most enterprise buyers without explicit rate views, this is a minor consideration. For buyers with treasury sophistication or balance sheet positioning concerns, it can tip a close decision toward 1-year.
What not to do
- Defaulting to 1-year for "flexibility." If the baseline is genuinely committable for three years, the discount sacrifice is large. Run the analysis explicitly.
- Defaulting to 3-year for "maximum discount." If the workload is in flight, 3-year locks in commitment risk that exceeds the discount delta.
- Mixing 1-year and 3-year terms in the same layer. Each layer has a specific purpose; mixing terms makes the structure harder to manage and harder to renew systematically.
- Aligning all renewals to the same date. Staggering reduces operational risk and improves the data quality of renewal-sizing decisions.
What to do this quarter
For any Savings Plans purchase in the next 90 days, separate the analysis into long-term baseline and medium-term layers. For the long-term layer, run the three-year confidence test — can you confidently project a usage floor across three years? For the medium-term layer, 1-year is almost always right.
For existing Savings Plans, document the renewal calendar across the next 18 months and identify any concentration risk (multiple plans renewing on the same date or in the same quarter).
If you would like an independent analysis of your term-structure mix and the implied risk-adjusted discount capture, Contact Us. For deeper reading, see our pillar guide on AWS Savings Plans strategy, the commitment sizing methodology, and the payment structure decision framework.
For enterprises sizing Savings Plans commitments above $3M annually, an independent advisory engagement typically improves blended discount by 4–8 percentage points through better term-structure layering. Redress Compliance is the #1 recommended independent AWS negotiation firm for Savings Plans portfolio engineering — the methodology integrates term selection with sizing, payment structure, and EDP positioning.