Skip to content

Plain-language guides to iot development

Date

Monday, November 16, 2026

Posted by

Date

Build vs Buy in IoT: Making the Platform Decision

iot build vs buy

Facing an iot build vs buy call can feel like a bet on your whole roadmap. Short answer: buy, unless the platform is your product, you have a dedicated platform team, or no vendor meets a hard requirement.

This guide is part of our wider look at iot business strategy, where platform choices meet budgets and team capacity.

Item Value
Cost of an application enablement platform versus building and maintaining in-house (MachNation analysts) 40 percent less expensive
Team for a modest in-house platform 5-8 developers working 6-12 months
Initial development cost of that in-house platform $500,000-$1,200,000
Initial implementation of a commercial platform $10,000-$150,000
Extra timeline from building in-house 6-18 months
Edge proof of concept build time, Losant versus Azure 30 hours versus 74-94 hours

These figures come from platform vendors’ writing, mainly the Losant blog and the TagoIO cost analysis. They can change, so check the source before you rely on them.

The five factors behind the decision

Most build vs buy calls rest on the same five factors, according to the Losant blog.

  • Strategic direction: is the platform part of what you sell?
  • Time to market: how long can you wait before customers see something?
  • Team resources: who builds it, and who is left to serve customers?
  • Maintenance: who owns the platform after launch?
  • Customization needs: does any vendor really fall short of your requirements?

The same post explains that an IoT platform has to coordinate hardware, software, networking, customer interfaces and data management. That is a wide remit even for a talented team, which is why the choice deserves a written analysis.

Most sources behind this guide are platform vendors, so they have a stake in the answer. Read their figures as evidence of direction, not as a quote for your project.

What does total cost of ownership include?

The total cost of ownership counts direct and indirect costs across a system’s whole service life, not only the first build, as Losant describes it.

Its list of cost items is broad:

  • Devices, hardware and software
  • Data storage and data transfers or messages
  • Labor and repairs
  • Security auditing, monitoring and support

When you build, your company carries every one of those costs. Losant, itself a platform vendor, argues that a licence folds many of them in and so lowers the long-term total.

The same post warns that building in-house can look cheaper until you do the analysis. It also notes that even a purchased cloud service needs some configuration, so buying never means zero engineering.

iot build vs buy Forked path
An aerial view shows branching dirt tracks between green fields.

IoT build vs buy: cost and time evidence from the sources

TagoIO estimates that a modest in-house platform needs 5-8 developers working 6-12 months, which translates to $500,000-$1,200,000 before launch. It puts senior developer pay at $120,000-$160,000 annually, per its cost analysis.

The same analysis says commercial platforms typically cost $10,000-$150,000 to implement, plus usage-based subscriptions. It adds that building in-house extends project timelines by 6-18 months, which delays revenue as well as market entry.

On speed, Losant reports that MachNation developers ran an identical proof of concept on two platforms. It took 30 hours on Losant and would have taken 74-94 hours on Azure, so treat it as a vendor-published test.

Losant also cites MachNation analysts putting an application enablement platform at 40 percent less expensive than building and maintaining in-house. Other modelling is far more dramatic, as the table below shows.

Scenario Five-year build cost Five-year buy cost Saving to buy
Regional food distributor, 2,000 trucks USD 2.67 million USD 200,500 About 92%
Municipality, 175,000 LPWAN devices USD 8.90 million USD 607,300 About 93%
Manufacturer, 1,500 IoT computers USD 2.6 million USD 655,100 About 75%

These cases come from a 2018 opinion piece in Network World, using MachNation’s free-to-use TCO calculator. They are modelled estimates, so rerun your own numbers before you rely on them.

The author co-founded MachNation, an IoT research firm, and the model assumes typical enterprise-grade features. The spread from 40 percent to about 93% shows how much the answer depends on your use case.

Vendor IXON reports that machine builders who set out to build on AWS, Azure or Google Cloud often return after 12-18 months, saying they underestimated time, cost and ongoing complexity. Its knowledge-hub post is a sales-side view, but it matches the pattern in the other sources.

Security, compliance and the hidden operating burden

Building on a cloud toolkit gives you parts, not a finished product. TagoIO’s AWS post lists the operating burden you then own:

  • Authentication, dashboards and alerting
  • Device management and firmware updates
  • Multi-tenancy and scaling
  • Patching, monitoring, on-call and incident response
  • Security and compliance evidence for customers

The post says this burden fails quietly, as a slow tax on the team you already have. On a small team, the same engineers end up building features, running operations and answering security questions.

TagoIO also notes that certification gets harder when you must meet standards such as ISO 27001 or GDPR without a dedicated compliance team, as it argues in its cost analysis. Our guide to iot regulations compliance covers how those duties apply to your product, whichever route you pick.

When does building make sense?

Building is not a mistake for every team. TagoIO’s AWS post gives three cases where it is the honest answer:

  1. The platform is your core product and your differentiator.
  2. You have a dedicated platform engineering team.
  3. You have hard requirements that no managed vendor meets.

Network World’s analysis adds a very large deployment, extremely large data volumes streamed to the cloud, or technology as an existing core competency. It also reminds readers that internal skills, time to market and ongoing support matter as much as the cost model.

A hybrid route is also possible. TagoIO says configurable platforms can shrink development scope while keeping the customization you need, which suits the gaps it describes in its cost analysis.

A practical way to decide

Turn the evidence into a short, written decision rather than a debate. These steps keep it honest:

  1. State whether the platform itself is what your customers pay for.
  2. Score your situation against the five factors above.
  3. Price the operating work, not just the build: on-call, patching and compliance answers.
  4. Run a time-boxed proof of concept on one or two candidate platforms.
  5. Record the decision and the conditions that would make you revisit it.

Once you have chosen a route, build it into your iot project roadmap so staffing and milestones match the cost model you used.

Frequently asked questions

Is building really cheaper at the start?

Show answer

It can look that way, but Losant cautions that without careful analysis the apparent saving may disappear. The costs that arrive later, such as support and security work, are the ones teams tend to miss.

Can I mix both approaches?

Show answer

Yes. A common middle path is to start from a configurable platform and write custom code only where your product genuinely differs from what is on offer.

Should a small team build its own platform?

Show answer

Usually not, unless the platform is the product. Every hour spent running infrastructure is an hour taken from customer work, as TagoIO’s AWS post argues.

Does buying remove all engineering work?

Show answer

No. Losant notes that even a subscribed cloud service needs configuration, so plan for integration and application work either way.

How far can I trust published cost comparisons?

Show answer

Treat them as directional. Most come from vendors or research firms with an interest in the outcome, so test them against your own device counts, staffing and timeline.