The case for NGS outsourcing is usually framed as a cost decision, and the case against it is usually framed as a turnaround decision: outsourcing is cheaper, owning is faster. Both halves of that framing are frequently wrong, and the second half is wrong in a way that catches labs by surprise. At the volumes where a lab is genuinely undecided about whether to own an instrument, owning it can actually deliver slower turnaround than a good commercial provider, because a lightly used instrument cannot fill a flow cell fast enough to run economically.
This article lays out a decision framework across three models, not two: running sequencing on your own instrument, using an institutional core facility, and contracting a commercial service provider. It builds directly on the cost break-even model in the Section 1 overview on building an NGS program and the real per-genome figures in how much NGS actually costs, and it assumes you have already worked through the readiness assessment, since a lab that is not ready to own well should be comparing a core facility against a commercial provider, not against itself.
Key Takeaways
|
Three Models and What Each Costs
Most in-house-versus-outsourced discussions quietly drop the middle option. An institutional core facility is neither fully in-house nor fully outsourced: you do not carry the capital or the service contract, but you often get faster turnaround, tighter sample control, and easier collaboration than a commercial provider offers, frequently at a subsidized internal rate. For a lab at a university or hospital that has a core, the real question is often core facility versus commercial provider, with fully in-house ownership a distant third that only makes sense at high sustained volume.
Model | What You Pay | What You Control | Best Fit |
Own instrument | Capital, service, staff, storage, all fixed costs regardless of volume | Turnaround, method, sample custody, scheduling | High, sustained volume with a real need for control |
Core facility | Per-sample internal rate, often subsidized; no capital | Some scheduling and method input; sample stays in-institution | Most academic and hospital labs, most of the time |
Commercial provider | Per-sample market rate; no capital, no staff | Least, but scales instantly and carries no fixed cost | Variable or low volume, or capacity overflow |
Table 1. Three sequencing models by what you pay, what you control, and where each fits best. The core facility middle option is the one most in-house-versus-outsourced comparisons omit.
Where the Break-Even Sits
The cost break-even between owning and outsourcing is not a fixed number, and any article that gives you one without stating its assumptions is guessing. It is governed by a single equation: owning becomes cheaper per genome only once your annual volume is high enough to dilute the fixed cost of the instrument below the gap between your variable cost and the outsourced rate. The Section 1 overview works this through in full; the short version is that the break-even volume ranges from roughly 440 genomes a year against a high outsourced rate to roughly 2,870 a year against a low one, using the cluster’s standard cost model.
The single most important consequence is worth restating plainly: the number that decides this is not your instrument quote. It is the outsourced rate you are comparing against, and that rate deserves multiple real quotes with a clear specification of what each includes. A lab that assumes a low outsourced rate will talk itself out of owning; a lab that assumes a high one will talk itself into a purchase that a real quote would have prevented. Get the real number before the model means anything.
Turnaround Time as the Deciding Variable
Here is where the framing most labs start with quietly falls apart. The assumption is that owning an instrument means faster turnaround, because the sample never leaves the building. That is true only if the instrument runs often enough to sequence your sample promptly after it is ready. At low volume, it does not, because sequencing runs are economical only when the flow cell is reasonably full, and filling a flow cell at low volume means waiting for enough samples to accumulate.
Annual Volume | Time to Fill a 16-Sample Flow Cell | Turnaround Implication |
150 genomes/yr | ~39 days | In-house batching wait alone likely exceeds a commercial provider’s total turnaround |
300 genomes/yr | ~19 days | In-house roughly matches a typical provider once shipping is added |
600 genomes/yr | ~10 days | In-house begins to win on turnaround |
1,200 genomes/yr | ~5 days | In-house clearly faster; the speed argument for owning now holds |
Table 2. Time to accumulate enough samples to fill a 16-sample flow cell at four annual volumes, assuming even sample arrival. Illustrative; real arrival is lumpy, which makes low-volume wait times worse, not better. Excludes the sequencing run itself and downstream analysis.
The Speed Argument for Owning Inverts at Low Volume At 150 genomes a year, a sample can wait over a month just for enough companions to justify starting a run, before the run itself even begins. A commercial provider batching across dozens of clients runs full flow cells constantly and can often return data faster than a lightly used in-house instrument can start. The lab that bought its own sequencer specifically for faster turnaround has, at that volume, bought slower turnaround, unless it chooses to run under-filled flow cells and absorb the much higher per-sample cost that comes with them. Owning for speed only delivers speed above the volume where flow cells fill quickly on their own. |
This is the point worth holding onto: the break-even is not a sample count, it is a sample count at a given turnaround requirement. A lab that needs three-day turnaround has a completely different decision from one that can tolerate three weeks, even at identical volume. Define the turnaround requirement first, then find the volume that satisfies it, and only then compare that combined figure against what each of the three models can actually deliver. A core facility, notably, often solves the low-volume turnaround problem better than either owning or a distant commercial provider, because it batches your samples with everyone else’s at the same institution and runs frequently.
Interested in lab tools and techniques?
Data Ownership and Confidentiality
Cost and turnaround are the variables that get modeled. Data custody is the one that overrides the model when it applies. Sending samples and the resulting sequence data to an external provider raises questions that never arise with in-house or, usually, core facility work: who owns the raw data, how long the provider retains it, whether it is used to improve the provider’s own reference datasets, where it is physically stored, and what happens to it if the provider is acquired or ceases trading.
For routine research on non-sensitive samples, these questions usually have acceptable answers and the cost and turnaround math governs. For work involving patient-identifiable data, controlled-access samples, proprietary strains or constructs, or anything with intellectual property implications, data custody can and often should override the economics entirely. A service agreement that does not clearly answer the ownership, retention, and deletion questions is not a cheaper option; it is an unpriced risk. The specific terms a sequencing service agreement should cover, and the comparison across commercial providers, see LabX coverage of How to Buy a DNA Sequencer: A Step-by-Step Process, for a detailed comparison.
Hybrid Models
The three models are not mutually exclusive, and the strongest programs frequently run two of them at once. The most common productive hybrid is owning an instrument sized for baseline committed demand while outsourcing overflow during peaks, which keeps the owned instrument well utilized, and therefore cheap per sample, without forcing a second capital purchase to cover demand spikes that may not persist. This is the pattern that lets a program grow into a larger instrument gradually rather than betting on projected volume up front.
A second useful hybrid runs routine, high-volume work in-house or at a core facility while sending specialized, low-volume assays to a commercial provider that already has the method validated. Building and validating a rarely used protocol in-house is expensive and slow; renting someone else’s validated version for the handful of samples that need it is often the better call, and it keeps the in-house team focused on the workhorse applications that justify the instrument in the first place. The overarching decision, whether to build any in-house capacity at all and at what scale, is covered in the main guide to next-generation sequencing in the lab. This article’s job is narrower: to make sure that when you compare the options, you compare all three, and you compare them on turnaround at your real volume, not on cost alone.
This article was produced under Lab Manager's AI Editorial Guidelines.
