# BYOC Build vs Buy: A Three-Year Cost Comparison

> Building BYOC in-house or buying it? This post compares the cost of both, for a vendor starting fresh and for one already running a home-spun BYOC and deciding whether to switch. Every assumption is listed, so you can test your own.

Mark Milligan · 2025-07-23
Source: https://nuon.co/blog/part-1-build-vs-buy-byoc-the-nuon-advantage-in-tco-and-roi

---

Customers asking software vendors to run their product inside the customer's own cloud account is not new. Most vendors who say yes build it themselves. The work does not shrink after the first few installs. It grows with every customer, every release and every cloud.

This post puts numbers to the build vs buy decision. In the example we model, a vendor adding two BYOC customers a month for three years, growing to 72 customers, spends about $1.3 million building and running BYOC in-house, against about $530,000 with Nuon. Across the cases we tested, in-house cost roughly 1.5 to 4 times as much. The gap comes mostly from people's time, not software, so it rests on assumptions you can check, and we list every one of them.

If you already run a home-spun BYOC, the question is different: switch or stay. We cover that case too, including the month the switch pays for itself.

Updated October 2026.

## Two Situations, and What Most Vendors Actually Build

The economics differ depending on where you are starting.

**Build vs buy.** You have not built anything yet and are deciding between building BYOC and adopting a platform. Everything counts: the build, the tooling and the people who run it.

**Switch vs stay.** You already run a home-spun BYOC and are deciding whether to move. Your build is paid for, so we treat it as sunk and compare only what happens from here: the cost of moving your existing customers over against the cost of staying put.

![](https://cdn.sanity.io/images/3jtsca34/production/95b5ae6a6ea76b78877cf27b938e08a250f48291-2898x1554.png?w=1200)

The in-house version tends to look similar from one vendor to the next. A team writes Terraform for one cloud, usually AWS, and sets up cross-account access so the vendor can reach the customer's account. The customer gets a document explaining how to prepare their side. From there, installs, troubleshooting and upgrades are done by hand by the vendor's engineers.

That works for the first few customers. Multi-cloud is usually left out on purpose. A second cloud means new Terraform, a new permissions model and a new set of quirks to learn and support, and the team that built the first one is rarely staffed to do it again.

So the in-house case here is that kind of build, and cloud count is an input. When we add clouds, we charge the in-house side for building and owning them too, so the comparison stays fair.

## What Makes Up an In-house BYOC

Running your application inside a customer's cloud account is more than deploying it there. A working offering has to cover:

- **Packaging.** The application as something a customer's account can install, such as container images and Helm charts for Kubernetes. A self-hosted vendor often has this already. A SaaS vendor usually does not, and producing an installable release, including the underlying infrastructure and database setup, is significant work.

- **Base infrastructure.** The Kubernetes cluster or equivalent that the application runs on, in each customer's account.

- **Access and permissions.** How your team reaches the customer's account without standing broad access, and what you are allowed to do once you are there.

- **Networking.** DNS, subnets and the connectivity the application needs.

- **Installation.** Provisioning all of it in a new customer account, repeatably.

- **Secrets.** Creating, storing and rotating credentials.

- **Day-2 operations.** Health checks, monitoring, drift between what you intended and what is running, and troubleshooting.

- **Upgrades.** Rolling every release out to every customer safely, and rolling back when something goes wrong.

- **Removal.** De-provisioning when a customer leaves.

Now multiply it. Each cloud you support has its own version of all of this, and each customer may need more than one environment, such as development, test, staging and production.

![](https://cdn.sanity.io/images/3jtsca34/production/152e9f7f9a5a853a95b2e9249f2c45de4bba9353-2898x1554.png?w=1200)

Even a well-run in-house build tends to stop short of a few things: more than one cloud, an audit trail of what your team did in each customer's account, automated upgrades across the whole customer base, policy checks and drift detection. Customers' security teams ask for these. We come back to them below.

## How We Model It

We model three years month by month. Cost on both sides grows with the number of customers, how often you release, and how many clouds you support, because that is how the work actually grows.

Assume an engineer costs $220,000 fully loaded, or about $106 an hour. Here is what each side spends time on:

![](https://cdn.sanity.io/images/3jtsca34/production/99f59b6f0af29dad0edaa638a5fe5e4485a6e8a6-2898x1554.png?w=1200)

The Nuon-side time is our estimate of the part-time effort a small team spends on installs, upgrades and support. The in-house numbers are our estimates too, not measurements, and they matter most. We test them below.

## The Example

Assume a vendor with no BYOC customers today, adding two a month for three years, so 72 customers by the end. One cloud, one release a month, $220,000 engineers, and Nuon set up with standard support.

![](https://cdn.sanity.io/images/3jtsca34/production/5dbcfa9868a4e7717aaf273f1482b3b5f4af3a4a-2898x1554.png?w=1200)

In-house, the three-year cost is about $1.33 million. Platform engineering is the largest share at 43%, followed by support and troubleshooting at 26% and releases at 16%. The initial build is 5%.

With Nuon, the three-year cost is about $528K, of which about $14K is setup. Most of the rest is still your own people's time, because Nuon takes over the work, not the customer relationship.

The difference is about $805K, roughly 3.7 engineer-years. By year three, running BYOC in-house costs about $595K a year, or 2.7 engineers, to serve 72 customers. The same year with Nuon costs about $223K.

Notice the build itself is only 5% of the three-year cost. People tend to think of the build as the expense. In this model, running it is.

### Switch vs stay

Now assume the same vendor already runs BYOC for 20 customers and is adding two a month. The build is paid for, so we compare only what happens from here. Moving 20 customers at one day each, plus the Nuon setup, costs about $30K once. Over three years, staying costs about $1.70M and switching about $669K. The switch pays for itself in the third month, because the in-house run cost is higher every month after. If each move takes three days instead of one, it pays back by month 5.

![](https://cdn.sanity.io/images/3jtsca34/production/7d15cf4874ce715f5a9716d7bd4fc2f2538a5b6e-2898x1554.png?w=1200)

## What Moves the Answer

Here is how the $805K difference changes when we change one thing at a time.

![](https://cdn.sanity.io/images/3jtsca34/production/e0a2d0343571839d069d12df2f456ce1e0576a47-2898x1554.png?w=1200)

Growth and release cadence matter most on the in-house side, because every customer and every release adds work that does not go away. On the Nuon side, the effort your own team spends is the assumption to watch.

**Where in-house gets close, or wins.** If your team is much leaner than our assumptions, in-house can tie. At about 40% of our in-house effort across the board, the two cost the same in this example. If your in-house effort is half of ours and your effort with Nuon is 1.5 times ours, in-house comes out about $60K cheaper. Small, flat fleets are the other close case: with 5 customers and no growth, the difference is about $155K, and switching pays back in month 4. The gap is wide in most of the cases we tested, but it depends on these assumptions, so test yours.

## Day-2 Capabilities You Would Also Build

Customers' security teams increasingly ask for a handful of capabilities beyond installing and upgrading. In-house, each is a one-time build. With Nuon they are included.

![](https://cdn.sanity.io/images/3jtsca34/production/f0cf6a144830eee62b4012308eaceac4eddd12f7-2898x1554.png?w=1200)

That is 23 engineer-weeks, about $97K at $220,000, which would take the in-house example from $1.33M to about $1.43M. We model these as a one-time build with no ongoing upkeep, which understates in-house cost.

## What This Leaves Out

These examples are a simplified look at cost, built to show the difference between building in-house and adopting a platform like Nuon. They leave out several things that often matter more:

- **Deals.** Vendors lose or delay customers who require BYOC. We do not put a value on that here.

- **Time to market.** An in-house build has to exist before the first customer installs.

- **Compliance work,** such as SOC 2 or HIPAA, which adds effort on both sides.

- **Opportunity cost.** The engineers in the in-house column are not building your product.

- **Pricing.** Nuon pricing depends on your customer base and your agreement, so we show Nuon's cost combined, not as a rate card.

The model works for fleets of up to a couple hundred existing customers and up to about 20 new ones a month. Beyond that, the numbers need a conversation, not a table. Please run your own numbers. Every business has different salaries, release habits and customers, and the assumptions above are illustrative.

## Six Numbers to Gather Before You Decide

- Customers on BYOC today.

- New customers a month.

- Clouds you need to support.

- Releases a month.

- Your loaded engineer cost, divided by 2,080 for an hourly rate.

- How much help you want from the vendor. 

![](https://cdn.sanity.io/images/3jtsca34/production/9240f705a69ba33b07c8c170cc1c4ffeeb61ff75-2898x1554.png?w=1200)

## Conclusion

An in-house BYOC build is a smaller expense than it looks, and an in-house BYOC operation is a larger one. In our example the build was about 5% of the three-year cost. Running it, which means onboarding customers, shipping releases, supporting installs and keeping a platform team, was the rest, and it grows with every customer, release and cloud you add. For a vendor already running a home-spun BYOC, the question is whether to switch, and the move can pay for itself in a few months. The larger cost either way is the engineering time an in-house build keeps off your core product. That is the time Nuon gives back, by making customer deployments repeatable and not a standing infrastructure project.

Please check out our [BYOC Cost Calculator.](https://nuon.co/tools/byoc-cost-calculator-lite)