Blog
1-Year vs 3-Year Committed Use Discounts
Google Cloud prices committed use discounts on a simple ladder: a 3-year CUD costs less per unit than a 1-year CUD, which costs less than on-demand. If cheaper is better, why not put everything on a 3-year term? Because the price is only half the decision. The other half is risk.
The discount is real — and so is the lock-in
A commitment is a promise to pay for a fixed level of capacity for the whole term, whether or not you end up using it. A 3-year term deepens the discount precisely because you're taking on more of that risk: you're betting that demand will still be there in year three. For stable, baseline demand you're confident about, that bet is easy and the deeper discount is free money. For demand that might churn, shrink, or move, the same bet can leave you paying for capacity you no longer need.
Match the term to the durability of the demand
The useful way to think about it is to slice your demand by how durable it is:
- Rock-solid baseline— demand you'd bet on for three years. This is where 3-year commitments earn their deeper discount.
- Probable but uncertain — demand likely to persist a year or two but not clearly beyond. 1-year commitments capture most of the discount with far less exposure.
- Volatile or short-lived — spiky or uncertain demand is often cheapest left on-demand, even at the full rate, because a wasted commitment costs more than the on-demand premium.
Break-even makes it concrete
The line between “commit” and “don't” is the break-even point: how many months a commitment has to stay utilized before it costs less than paying on-demand for the same usage. A 3-year term has a longer break-even than a 1-year term, so it needs more confidence in the out-years to pay off. If a slice of demand is likely to disappear before its commitment breaks even, that slice shouldn't be committed — or should take a shorter term.
Why this is an optimization problem
In practice your demand isn't one clean slice — it's many contracts starting and ending at different times, partially covered by commitments you already own, evolving month by month. The best split of 1- and 3-year commitments across all of it is exactly the kind of coupled, over-time decision a mixed-integer program is built to solve. That's what the Solvicus CUD optimizer does.