Build or buy? Run the five-year numbers first.
Put your seat count, SaaS quote and growth plans next to the real cost of owning custom software. The model counts licence rises, AI add-ons, maintenance, hosting and internal time, and shows how an AI-native build team changes the break-even point.
Compare five years of ownership
Pick a starting point, then replace the prices with your own quotes. Every figure updates as you go.
- Buy SaaS$450,000
- Custom build$510,000
- Hybrid$440,000
- Year 1$55,000$260,000$130,000
- Year 2$120,000$320,000$200,000
- Year 3$210,000$380,000$270,000
- Year 4$320,000$450,000$350,000
- Year 5$450,000$510,000$440,000
- Licences$170,000
- AI add-ons and usage$35,000
- Build, setup and switching$77,000
- Maintenance$68,000
- Hosting$16,000
- Internal time$71,000
- Price rises of 7% a year add about $43,000 to SaaS licences over five years. Ask for a multi-year cap before you sign.
- Seats grow from 60 to about 124 by year five, so per-seat pricing compounds while custom hosting grows far more slowly.
- AI add-ons make up $130,000 of the SaaS five-year cost. In a custom build you pay for AI usage directly and control which data it sees.
- The custom build does not pay back against SaaS within five years under these assumptions.
- The hybrid breaks even with SaaS after 4.7 years, with a smaller build to own and maintain than a full custom system.
- Compared with a traditional team, an AI-native build costs about $52,000 less up front and goes live roughly 1.0 months sooner under these assumptions.
- Break-even is unchanged (beyond 5 years), but the lower build cost cuts the five-year custom total by about $75,000.
- AI lowers the cost of writing code, not of owning it: hosting, security and an accountable owner remain.
Indicative 2026 assumptions. The hybrid assumes a lower SaaS tier at 60% of the seat price plus a custom layer at 40% of the full build. Maintenance is a share of the traditional build cost, reduced for AI-assisted upkeep.
Indicative only. We will send the inputs above with your message so a senior engineer can sanity-check them.
A year-by-year total cost of ownership, not a sticker price.
For each of the five years, the calculator works out what you would pay under three options. Buying SaaS means per-seat licences that grow with your headcount and the vendor's annual price increases, plus any AI add-on, implementation and the time someone spends administering the product. Building means an upfront build, then maintenance, hosting, AI usage and the internal time needed to own a product roadmap.
The hybrid option keeps a lower SaaS tier for commodity functions and builds a thin custom layer on top for the parts that make you different. That is often where the best answer sits, so it is worth seeing its numbers even if you arrived expecting a straight choice.
Build cost starts from typical effort for the complexity you pick, adjusted for how the team uses AI, then converted at your blended rate. The build also takes time: if you already run a SaaS product, you keep paying for it until the custom system is live, and the model includes that overlap. Break-even is the point where the cumulative cost of building drops below the cumulative cost of buying.
What goes into software total cost of ownership
Most build vs buy spreadsheets miss at least two of these lines. Each one is an input you can change in the calculator above.
| Buy SaaS | Custom build | Hybrid | |
|---|---|---|---|
| Year one | Implementation, integration, data migration | Build effort, switching, overlap with current tools | Smaller build plus a lighter implementation |
| Recurring fees | Per-seat licences that rise every year | Hosting that grows slowly with users | Lower-tier licences plus modest hosting |
| AI costs | Per-seat AI add-ons on top of licences | Model usage you pay for and control | Usage costs on the custom layer only |
| Upkeep | Included in the licence | Commonly 15 to 25% of the build cost a year | The same share, on a smaller build |
| People | An admin configuring and supporting it | A product owner setting priorities | Some of both |
| Exit risk | Lock-in, export limits, price changes | Key-person risk if nobody documents it | Split across two, usually manageable |
Maintenance percentages are typical industry rules of thumb, not guarantees. Complex or regulated systems sit at the higher end.
AI moved the break-even point. It did not remove the cost of owning software.
AI-native engineering teams run coding agents on integrations, standard screens, tests and documentation, with senior engineers designing the system and reviewing every change. For many business applications that means fewer engineer-hours for the same scope, so the upfront cost of building falls and the build goes live sooner. Switch the delivery approach in the calculator and watch the break-even year move.
At the same time, SaaS vendors are adding AI features and pricing them as per-seat add-ons, which compounds with headcount in the same way licences do. A custom build pays for model usage directly, which can be cheaper at scale and keeps you in control of what data the AI sees.
What AI does not change is accountability. Someone still has to patch dependencies, watch security, keep hosting healthy and decide what to build next. If nobody in your organisation can own that, the cheaper build is a false saving.
Six ways to make the build vs buy numbers honest
Use a real quote, not the pricing page
Enterprise tiers, minimum seat counts and paid support often double the published per-seat price.
Model headcount growth you actually expect
Per-seat pricing punishes growth. Use your hiring plan, not today's team, and include contractors who will need access.
Ask how the build team uses AI
Two quotes for the same scope can assume very different delivery approaches. Ask what is automated, what is reviewed and whether savings reach you.
Price the exit as well as the entry
Check data export formats and contract notice periods now. Switching cost is the clearest measure of lock-in you will get.
Count internal time at its real cost
An operations manager spending a day a week administering a tool is a meaningful cost line whichever way you go.
Test the verdict with pessimistic inputs
Raise maintenance to 25% and slow the growth rate. If the answer flips, treat it as a close call and weigh control and risk instead.
Questions we often hear
How do you calculate total cost of ownership for software?
Add every cost over a fixed period, usually three to five years: licences or build cost, implementation, AI add-ons or usage, maintenance, hosting, internal staff time and the cost of switching. Compare the cumulative totals year by year rather than the first-year price, because per-seat fees and maintenance grow at very different rates.
Is SaaS cheaper than custom software?
In the first year, almost always. Over five years it depends on seat count, growth, price rises and how much of the product you actually use. Large or fast-growing teams on per-seat pricing often find a custom or hybrid build cheaper by year three or four, while small teams with standard needs rarely do.
How does AI-native development change build vs buy?
Teams that work AI-first typically need fewer engineer-hours for routine code, integrations and tests, so custom builds cost less and ship sooner, which pulls the break-even year forward. Novel logic and product decisions benefit less, and AI-written code still needs senior review, so the saving is real but not unlimited.
What maintenance percentage should I use for custom software?
A common rule of thumb is 15 to 25 percent of the original build cost per year, covering dependency updates, security patches, bug fixes and small improvements. Use the lower end for a simple internal tool with few integrations and the higher end for customer-facing or regulated systems.
When does a hybrid approach make the most sense?
A hybrid works well when most of the capability is standard but one workflow, portal or AI feature sets you apart. You keep a proven product for the commodity parts on a cheaper tier and build only the differentiating layer, which limits both licence growth and the amount of software you have to maintain.
Related reading and tools
Want a second pair of eyes on this decision?
Send your calculator results and the products you are considering. We will give you a vendor-neutral view on whether to build, buy or combine the two, and what an AI-native build would realistically involve.
Working with companies globally · Response within 24 hours