A Prepayment Discount Is a Loan: How to Price the Implied Interest Rate Before You Sign

A prepayment discount is not a discount. It is an interest rate, and almost nobody prices it as one.

When a host offers "two months free on an annual plan," the invoice language suggests a price cut. What has actually happened is that you agreed to hand the vendor ten months of fees today in return for twelve months of service, which means you lent that vendor your cash. The "discount" is the return on that loan. The moment you accept that framing, three questions become answerable in a way they never were while you were reading the word free: what interest rate is the vendor paying you for the use of your money, how does that rate compare with what the same cash earns doing nothing at all, and how long do you have to stay for the loan to beat the alternative.

This article does that arithmetic from first principles. It is not a guide to any one promotion. It is a method for reading any prepayment offer, including the ones that do not advertise a percentage at all.

The Two Kinds of Cloud Discount Are Not the Same Instrument

The reason prepayment is mispriced so often is that the cloud industry uses one word, discount, for two instruments with completely different risk profiles. Sorting them apart is the whole of step one.

The distinction between the two is what makes the hourly-versus-monthly question a separate decision from the prepayment question, and the workloads that genuinely save on hourly billing are exactly the ones for which prepayment is structurally wrong, because their run pattern is intermittent by design rather than by exception.

A commitment discount is a price concession you receive in exchange for promising volume. You pay nothing up front. You commit to consuming at least a certain amount over a term, and the unit price drops for the whole term. AWS publishes this shape plainly: Standard Reserved Instances carry an average discount of 40 percent for a one-year term and 60 percent for a three-year term, with a stated maximum of 72 percent; Convertible Reserved Instances average 31 percent and 54 percent with a stated maximum of 66 percent. Google Cloud's committed use discounts for Cloud SQL publish a flat 25 percent for a one-year commitment and 52 percent for a three-year commitment, with overage billed at the on-demand rate. In both cases there is no cash outlay. You are not lending anyone anything. You are accepting an obligation to buy, and the vendor is paying you for the certainty of that revenue.

A prepayment discount is a different animal. You hand over money before you receive the service. The vendor now holds your cash and owes you a service. That is a loan, with a term, a principal, and an implied rate, whether or not anyone writes those words on the page.

Both instruments can produce the same headline percentage, and the percentages are not comparable. A 25 percent commitment discount costs you nothing and carries only the risk that you stop using the service. A 25 percent prepayment discount requires you to fund a year or more of spend today and carries every risk a lender carries. Treating them as the same number is the most expensive mistake in this whole area, and it is the default assumption for most buyers.

The Only Number That Makes Two Offers Comparable

To compare offers you must convert each one into a single quantity: the annualized rate of return on the cash you hand over. Everything else, including the percentage off list, including the number of free months, is an input to that figure rather than the figure itself.

The reason the headline percentage misleads is that the discount is normally quoted against the list price, while the cash you actually part with is the discounted amount. Those are different denominators. One month free on a twelve-month term is 8.33 percent off the list price, but you only ever paid 91.67 percent of list, so measured against the money that actually left your account the concession is 9.09 percent. Free months on a longer term widen the gap further: three months free on a twelve-month term is 25 percent off list and 33.33 percent off the amount paid. The percentage you are quoted is always the smaller of the two.

The annualized figure is larger still, because the discount is earned over the term rather than instantly. To compute it, solve for the rate at which the discounted sum you hand over today equals the same monthly fees paid across the term. In units of one month's fee, for a twelve-month term with one month free, you are solving for the rate r in:

11 = the sum, from month 1 to month 12, of 1 divided by (1 + r) to the power of the month number.

The solution is a monthly rate of about 1.365 percent, which compounds to roughly 17.7 percent annualized. That is the rate the vendor is implicitly paying for your cash, and it is the only number that can be compared against a bond, a money-market fund, or another host's offer.

From "Months Free" to an Annualized Return

The table below converts the wording you actually see on a pricing page into the figure you actually need. Every row assumes a twelve-month term and prepayment of the full discounted amount, which is the common shape of an annual plan.

Two things in that table deserve attention because they invert the intuitive reading. First, the annualized return is always roughly double the headline discount, and the gap widens as the discount grows. A "3 months free" offer reads like a quarter off; it is actually closer to three quarters of annualized return on the money you front. Second, the discount against the amount paid is systematically larger than the discount against list, which means any vendor quoting the list-based figure is quoting the flattering direction of the same deal.

Neither of those observations says the offer is bad. They say the offer is not the size the page claims, in either direction, and that you cannot judge it until you have done the division. The same normalisation habit is what a proper pricing benchmark across providers does at the instance level: the sticker is never the comparable quantity.

The Same Sentence Means Two Different Deals at Two Different Terms

Here is the trap that catches buyers who have learned to do the arithmetic but stop halfway. The phrase "three months free" has no intrinsic value. Its value depends entirely on the length of the term it sits inside, because you are tying up the money for that whole term.

Three months free inside a twelve-month term is 25 percent off list and produces the 74.1 percent annualized return in the table above. The identical offer, three months free inside a thirty-six-month term, is also 8.33 percent of the term, but the cash is locked for three years instead of one, and the annualized return collapses to about 5.9 percent. One month free on a twelve-month term returns 17.7 percent annualized; three months free on a thirty-six-month term returns 5.9 percent. The second offer sounds three times more generous and is worth a third as much per year.

The general rule that falls out of this is short and worth memorizing: a prepayment discount must be judged per year of lock-up, not per term. A discount that looks large because the term is long is often the worse instrument, and the vendors that price long commitments aggressively are counting on you comparing the number of free months rather than the annualized rate.

The same arithmetic runs the other way in the commitment-discount world, which is a useful sanity check. A forty percent commitment discount over three years, which requires no cash at all, has an implied annualized value of roughly 44 percent against the undiscounted path, and a 52 percent commitment discount over the same term implies about 73 percent. Those figures are not interest, because you never parted with the money. They are a measure of how much the vendor is willing to concede for certainty, and they are the reason a no-upfront commitment should almost always be evaluated before a prepayment is considered at all.

What the Risk-Free Rate Pays Instead

A rate of return is meaningless in isolation. The comparison that matters is against the rate you would earn on the same cash with no credit risk and no lock-up, which in US dollars means Treasury bills. The figures below are the coupon-equivalent yields published by the US Department of the Treasury on 21 October 2026, the most recent trading day at the time of writing.

Now the arithmetic bites, and it bites in the direction most buyers do not expect. To match a 4.49 percent one-year Treasury bill, a twelve-month prepayment offer needs to give you only about 2.34 percent off the list price, which is roughly 0.28 of a month free — about eight and a half days. To beat the Treasury bill by a factor of two, the hurdle is about 4.52 percent off list, or roughly sixteen and a half days.

That is a low hurdle, and the correct conclusion is the uncomfortable one: across the range of prepayment offers actually advertised, the discount clears the risk-free rate easily. An eight percent discount for annual prepayment is not a bad rate at all — it converts to about 16.9 percent annualized, roughly four times the one-year Treasury bill. A twenty-five percent discount for a year up front converts to about 74 percent annualized. Judged purely as a rate, the vendor is paying generously.

Which is precisely why the risk-free comparison is the wrong place to conclude the analysis, and the most common place people stop. The discount is not the problem. The problem is that the rate is paid in a currency you cannot spend: the return arrives as service you must consume, from a vendor that must remain solvent, in a configuration you are usually forbidden from changing, and it is forfeited in part if you leave early. The rate is high because the instrument is illiquid and unsecured, and the spread over a Treasury bill is the price the market puts on those two facts. The five risks in the next section are what you are being paid to accept, and the correct question is whether the spread is large enough to compensate you for them — not whether the discount exists.

One caveat, stated plainly because it affects the conclusion: this comparison is denominated in dollars against dollar rates. If your cash would otherwise sit in a currency with different rates, or if you are financing the prepayment on a credit line at a higher rate than the implied return, the hurdle moves. Anyone paying 20 percent annualized on a business credit card should almost never prepay a cloud bill, regardless of how attractive the discount looks, because the borrowing cost exceeds the implied return by an order of magnitude.

The Break-Even Utilization You Are Underwriting

A prepayment is a one-way bet on your own forecast. You have paid for the whole term up front, so the value of the offer depends on how much of the term you actually consume. The question is not "will I use the server this year" but "how many months must I keep it for the prepayment to have beaten paying monthly." The table below answers it by counting months directly, which is the honest way to do it: the prepaid sum buys a fixed number of months of service, and the monthly path costs one unit for every month the workload lives.

The shape of the damage is steeper than the table suggests, because the loss is not linear in the shortfall. If you take one month free and the workload dies after six months, you paid eleven months of fees for six months of service, which is 83.3 percent more than monthly billing would have cost you. If it dies after eight months, you are 37.5 percent worse off; at nine months, 22.2 percent. If it survives eleven months, you break even exactly. Only in the twelfth month does the discount begin to have been worth anything, and the reward for surviving the full term is an 8.3 percent saving against the monthly path. The asymmetry is the point: the downside is capped only by the size of the prepayment, while the upside is capped at the discount.

This is why the forecast quality matters more than the discount size. The industry data on forecast quality is not encouraging: Flexera's 2026 State of the Cloud Report, a survey of 753 cloud decision-makers across the fifteenth edition of the study, found that organisations estimate 29 percent of cloud spend is wasted, a figure that rose for the first time in five years and that the report attributes in part to AI workloads with usage patterns that are difficult to forecast. A prepayment is a bet that your own utilisation forecast is in the accurate 71 percent, and the base rate says otherwise for about a third of spend.

The practical version of this rule is to prepay only the portion of your footprint that is structurally constant. A production database that has run on the same specification for two years is a candidate. A staging environment is not, no matter how confident anyone is that it will still be needed in eleven months.

Five Risks the Discount Does Not Cover

The annualized return in the first table is a gross figure. It is what you earn before subtracting the risks that prepayment introduces and monthly billing does not. Each of the five below is a real cost that has to be netted against the implied rate before the offer can be called good.

Three of the five risks above are the same risks that decide the cost of leaving a provider altogether, which is why exit cost should be read as a contract term rather than as an afterthought: the clause you sign determines what the risks in this table are worth in dollars. When a provider states that early removal is refunded "based on the list price," that phrasing does not resolve the question by itself; it creates one. A refund basis that refers to the undiscounted price and a refund basis that refers to the discounted price you actually paid produce materially different outcomes for the same exit, and the difference is largest precisely when the discount was largest. Read the exit clause as a formula with an unknown in it, and ask the vendor to state the unknown.

A Four-Question Screen Before You Prepay

Run these in order. The screen is designed so that a no at any step ends the analysis, which saves the effort of computing returns for offers that were never going to win.

1. Is there a no-cash commitment offer available first? If the vendor offers a commitment discount without prepayment, evaluate that before the prepayment. You are comparing a free concession against a loan, and the free concession usually wins.
2. What is the implied annualized return, and how does the spread over Treasuries pay you for the lock-up? Convert the free months or the percentage off into an annualized figure using the method in the second section. The rate will almost always exceed the one-year Treasury bill of 4.49 percent — the useful question is whether the spread compensates you for illiquidity and vendor exposure, which is what the next two steps decide.
3. What is the break-even month, and is the workload structurally constant past it? If the workload is not certain to survive past the month count from the break-even table, the prepayment is a bet on your forecast rather than a discount.
4. What does an early exit return, and can the configuration change? Any offer that forbids resizing or moving the workload during the term carries a higher effective risk, and the implied return must be discounted accordingly.

A fifth check belongs here as a discipline rather than a step: write the implied rate down next to the offer. The number is the whole point of the exercise, and the habit of recording it is what turns this from an article into a practice.

Where the Arithmetic Points When the Workload Is Genuinely Steady

There is a legitimate case for prepaying, and it is narrower and more specific than the marketing suggests. It requires all three of the following to hold at once: the workload is structurally constant for the entire term rather than merely expected to be, the configuration will not need to change, and the implied annualized return clears the risk-free rate by an amount you would accept for taking on vendor credit and liquidity risk. When those three conditions hold together, the prepayment is genuinely the better instrument, and it is usually correct to take the longest term the vendor prices best, because a longer term with a proportionally larger discount can still win on annualized return even though the term is longer.

The reason the billing options on a plan are worth reading before the discount is that they determine which of the three conditions you can even satisfy. A host that offers hourly, monthly, and yearly billing on the same configuration gives you a way to test the workload on the monthly rate first and commit only once the utilisation pattern is observed rather than forecast. That sequencing, trial first and commitment second, removes the largest single risk in the analysis: the risk that your own forecast was wrong. The measurement discipline for deciding what to size the instance at before committing is its own exercise, and it is worth doing before the term starts rather than after.

This is the point at which a specific provider's structure becomes relevant, because the structure is what makes the arithmetic tractable. On the bundled compute plans SurferCloud offers through the promotions section, the published billing options run across hourly, monthly, and yearly, which means the same configuration can be used at any of the three granularities and the prepayment decision can be deferred until after a month of real data exists. The plans document a configuration range from 1 vCPU and 1 GB up to 64 vCPU and 512 GB with dedicated vCPU allocation, unlimited traffic, and dedicated bandwidth, alongside a lighter product line that trades scalability for a lower entry price with shared bandwidth and a 30 Mbps port. That spread matters for this analysis because it separates the two populations that the break-even table treats identically: the workloads whose specification is a settled fact, and the workloads that will need to grow.

Two published terms bear directly on the five risks above and should be read before any commitment rather than after. The first is that renewal pricing is documented as staying the same as the initial rate, which removes the renewal-cliff risk that makes some prepayment offers look better than they are when the second year is priced at full list. The second is the exit basis. The promotion terms state that the campaign does not support changing zones or configurations during the period, and that if resources are deleted midway a refund will be made based on the list price. The no-change term is the flexibility risk made concrete, and it is the condition that a growing workload will violate; the refund basis is the fourth question from the exit screen, and it is stated in the terms rather than left to inference. Where a promotional plan is marked as out of stock at the time of reading, that is an inventory fact rather than a pricing one, and it does not change the arithmetic.

For a workload that is genuinely constant, cannot change shape, and has cleared the four-question screen, the bundled plans are worth pricing against the month-by-month alternative, since the whole decision reduces to whether the implied annualized return beats what the cash would otherwise do. The bundled plan pricing and terms are published here, with the line items and the discount basis stated on the page rather than in a quote.

Frequently Asked Questions

Is a prepayment discount ever bad value? Yes, but rarely because the rate is too low, and usually because the lock-up is wrong for the workload. One month free on a twelve-month term implies about 17.7 percent annualized, and an 8 percent discount for annual prepayment implies about 16.9 percent — both far above the 4.49 percent one-year Treasury bill. The rate is almost always generous. What makes an offer bad value is a term that outlasts the workload's reasonable lifetime, an exit clause that forfeits the prepaid portion, or a specification that will need to change before the term ends.

Why is the annualized return higher than the headline discount? Because the headline is measured against the list price while you only ever paid the discounted amount, and because the return is earned across the term rather than received instantly. One month free is 8.33 percent off list, 9.09 percent off the amount paid, and 17.7 percent annualized. The three numbers describe one offer, and the last is the only one that compares across offers.

Should I prepay a thirty-six-month term if the discount is large? Only after converting to an annualized figure. Three months free on a thirty-six-month term is 5.9 percent annualized, against 74.1 percent for the same three free months on a twelve-month term. One month free on a thirty-six-month term is worth about 1.9 percent annualized. Term length destroys the value of a fixed free-month offer faster than most buyers expect, and the comparison must be per year of lock-up.

How does this compare with a reserved instance or a committed use discount? Those are commitment discounts, not prepayments, so there is no cash outlay and no implied interest rate to compute. They are also, for that reason, usually the better first option: AWS publishes Standard Reserved Instance averages of 40 percent for one year and 60 percent for three years, and Google Cloud publishes 25 percent and 52 percent for Cloud SQL committed use. When a no-cash commitment is available, evaluate it before any prepayment.

What if I might leave early? Then the prepayment is a bet on your forecast, and the break-even table sets the threshold. One month free requires surviving past eleven months to break even against monthly billing; below that you paid more than monthly would have cost. The refund basis in the exit clause, not the discount, determines how expensive being wrong actually is.

Does the currency matter? Yes. The comparison above is denominated in dollars against dollar rates. If your cash earns a different local rate, or if you would finance the prepayment at a rate above the implied return, the hurdle changes and the conclusion can reverse.

The Short Version

A prepayment discount is a loan, and the implied rate is the only number that makes offers comparable. Convert free months into an annualized return by solving for the rate at which the discounted sum you pay today equals the monthly fees across the term; one month free on a twelve-month term is 8.33 percent off list and 17.7 percent annualized, while the same three free months on a thirty-six-month term are worth 5.9 percent and one free month on that term only 1.9 percent. The rate itself is rarely the problem — matching the 4.49 percent one-year Treasury bill of 21 October 2026 takes only 2.34 percent off list, about eight and a half days of service — so the decision is not whether the discount is large enough but whether it compensates you for the illiquidity and the vendor exposure. Compute the break-even month, because one month free does not break even until month eleven, and being six months wrong costs 83.3 percent against paying monthly. Then net the five risks the discount does not cover, of which the exit refund basis and the no-change clause are the two that decide the outcome. When a no-cash commitment discount is available, take it first. And prepay only the portion of the footprint that is structurally constant rather than merely expected to be, because at a 29 percent self-reported waste rate, the forecast is the risk.