List-price math only. Real EC2 bills can be cut further by reserving across multi-account organizations (Standard RIs auto-share within an Org), using Savings Plans for non-EC2 compute (Fargate, Lambda), and right-sizing instances before reserving. The deep audit models your real Cost Explorer + RI Coverage report to rank reduction wins by dollar impact.
Cumulative spend month-by-month for On-Demand vs your chosen RI, including the upfront payment as month-0 cost. The break-even month is when cumulative RI cost falls below cumulative On-Demand cost.
| Line item | Math | Value |
|---|---|---|
| — | ||
| Instance | On-Demand | 1yr All Upfront | 1yr No Upfront | 3yr All Upfront | 3yr No Upfront |
|---|---|---|---|---|---|
| t4g.medium | $0.0336/hr | $0.0232/hr | $0.0242/hr | $0.0161/hr | $0.0175/hr |
| m5.large | $0.096/hr | $0.0662/hr | $0.0691/hr | $0.0461/hr | $0.0499/hr |
| m5.xlarge | $0.192/hr | $0.1325/hr | $0.1382/hr | $0.0922/hr | $0.0998/hr |
| c5.2xlarge | $0.34/hr | $0.2346/hr | $0.2448/hr | $0.1632/hr | $0.1768/hr |
| r5.2xlarge | $0.504/hr | $0.3478/hr | $0.3629/hr | $0.2419/hr | $0.2621/hr |
One-page checklist of the biggest EC2 levers — RI vs Savings Plans by workload type, right-sizing before reserving, Graviton migration ROI, Spot for stateless workloads, the multi-account RI sharing trap, Convertible RI exchange mechanics, the AZ-pinning capacity trade-off, and the four most common reasons teams over-reserve. PDF sent to your inbox.
monthly_cost = effective_hourly_rate × hours_per_month × quantity
Example: m5.large at $0.096/hr On-Demand × 730 hours/month × 1 instance = $70.08/month On-Demand. The same instance under a 1-year All Upfront Reserved Instance has an effective hourly rate of $0.0662/hr (31 percent discount), so $0.0662 × 730 × 1 = $48.33/month equivalent. The 1-year All Upfront RI requires paying $580 upfront on day one — that's the entire annual cost prepaid — which is how AWS funds the discount. Partial Upfront splits roughly half upfront + half across 12 monthly bills. No Upfront has zero day-one cost but spreads the entire (smaller) discount across monthly bills only.
The 3-year terms multiply the math by 36 months instead of 12. A 3-year All Upfront on m5.large is $1,512 upfront for $0.0461/hr effective ($33.65/month equivalent) — a 52 percent discount but a 3-year commitment to that exact instance family and AZ. Convertible RIs let you exchange families mid-term at the cost of a slightly smaller discount.
Savings Plans usually beat RIs on flexibility, not headline discount. Compute Savings Plans apply to any instance family, size, region, OS, and tenancy (including Fargate and Lambda), so one commitment covers a workload that shifts from c5.2xlarge to m6i.2xlarge to Fargate without losing the discount. EC2 Instance Savings Plans lock to one family in one region but allow size flexibility within that family, giving slightly better discounts. RIs are the right pick when you also want capacity reservation (Standard Zonal RIs reserve an AZ slot) or when you need Convertible RI exchange flexibility. Rule of thumb: if you only need the price discount, pick Compute Savings Plans. If you also need guaranteed capacity in a specific AZ, pick a Standard Zonal RI.
3-year All Upfront cuts about 52 percent vs On-Demand; 1-year No Upfront cuts about 28 percent. The 3-year is a bigger bet on three unknowns: that the workload still exists at year 3, that the instance family is still optimal, and that AWS doesn't release a generation that obsoletes the family. Historically AWS has done exactly that on every major transition. Pick 3-year All Upfront only when the workload is steady-state production on a current-generation family with no replacement coming. Pick 1-year No Upfront when usage is variable, mid-architecture-migration, or you want option-value to switch families. Cash flow matters too: 3-year All Upfront on a single m5.xlarge ties up about $1,000 day-one.
Standard RIs keep billing for the remainder of the term regardless of whether the instance runs. The discount applies to any matching instance in the linked-account organization. Three escape hatches: the Reserved Instance Marketplace lets you sell unused Standard RIs to other AWS customers (typically recovering 60-80 percent of remaining value); Convertible RIs can be exchanged for a different family at the same or higher equivalent value; or if any account in the Org runs a matching instance, the discount auto-applies. A 3-year RI committed to a workload killed at month 8 is mostly lost money with about 60 percent recoverable via the Marketplace.
The discount percentage is roughly constant across families — 28-31 percent for 1-year terms and 48-52 percent for 3-year terms. What changes with the family is the dollar magnitude and the cash commitment for All Upfront. A t4g.medium saves about $11/month per instance on 3-year All Upfront — meaningful at scale but easy to over-engineer. An r5.2xlarge saves about $191/month per instance — that's where RIs pay off fastest. Graviton-based families (t4g, m6g, c6g, r6g) get an additional baseline discount versus Intel/AMD equivalents because the underlying ARM chips are cheaper for AWS to operate. The rule: maximize RI coverage on the largest, most expensive instances first, then walk down by hourly cost. Skip RIs where absolute monthly savings are under $20.
No. Instance family, hours per month, term, payment option, and quantity all run locally in your browser. The page fires an anonymous pageview beacon and CTA-click events so we can measure whether the calculator is useful — no inputs, no email (unless you submit one to the cheat-sheet form), no IP stored raw.