How much does IoT development cost? (2026 breakdown)

A factory floor with 200 sensors feeding real-time temperature data into an alert dashboard. A fleet of 500 delivery vehicles tracked over cellular. A building management system controlling HVAC across 30 properties. These are production IoT systems, not prototypes.

What they cost to build varies by an order of magnitude. This post breaks it down: what you're actually paying for at each layer, what drives the range, and what the ongoing costs look like once you're in production.

Key Takeaways

• A production IoT system with 100–10,000 devices costs $50,000–$120,000 to build. Firmware complexity and connectivity type are the two biggest cost drivers.

• Custom firmware vs off-shelf SDK is a 2–4x cost difference. Custom embedded engineering starts at $30,000 before any cloud or app work.

• Cellular connectivity adds $2–$30/device/month in ongoing costs. At 1,000 devices on cellular, budget $2,000–$30,000/month before application costs.

• Managed platforms (AWS IoT Core, Azure IoT Hub) are the right choice for 90% of builds. Custom infrastructure only pays off above 100,000 active devices.

What you're actually paying for in an IoT build

IoT is not a single technology. It's a stack with six distinct layers, and the cost lives in all of them. Before any vendor gives you a number, they should be scoping each layer separately.

Device and hardware layer

If you're buying commercial off-shelf hardware (temperature sensors, GPS trackers, smart meters with standard SDKs), the device layer is minimal engineering work. If you need custom firmware (a microcontroller running C/C++ with custom logic), expect $10,000–$30,000 for basic firmware and $30,000–$80,000+ for complex embedded systems with real-time OS requirements.

Most "IoT cost" discussions focus on the cloud and application layers. In practice, the hardware-to-cloud bridge (device selection, SDK evaluation, and provisioning scripts) takes 2–4 weeks regardless of whether you're building custom firmware or not. That's the step that's almost always underestimated in initial quotes.

Connectivity layer

The protocol you choose has a direct impact on both build cost and ongoing operating cost:

• WiFi: cheapest to implement, free data connectivity. Works for fixed devices near a router. Limited range.

• Bluetooth / BLE: low power, short range. Suited for asset tracking in a warehouse or retail store.

• LoRaWAN: low power, long range, very low data throughput. Good for agricultural or utility sensors that send small payloads infrequently.

• Zigbee / Z-Wave: mesh networking, common in smart building and industrial automation. Requires a local hub.

• Cellular (LTE / 5G / NB-IoT): works anywhere, most expensive. Requires a SIM and data plan at $2–$30/device/month depending on data volume and carrier.

Protocol selection affects hardware cost (cellular modules cost $5–$30 more per unit), network infrastructure requirements, and ongoing data costs. This decision gets made early and is painful to reverse.

IoT gateway

Many deployments require a gateway: an edge device that aggregates data from local sensors and forwards it to the cloud. Gateways handle protocol translation (e.g., Zigbee to MQTT), local buffering when connectivity drops, and edge processing to reduce cloud data transfer costs. A gateway adds $5,000–$20,000 in firmware and software development depending on whether you're buying off-shelf hardware (Raspberry Pi, industrial gateway appliances) or building custom.

Cloud backend

The cloud layer is where the engineering cost is most predictable:

• Device registry: managing which devices exist, their current state, and configuration

• Message broker: MQTT or AMQP to receive telemetry from devices at scale

• Time-series database: storing sensor readings efficiently (InfluxDB, TimescaleDB, AWS Timestream)

• Data storage: raw event logs, historical data, device audit trail

AWS IoT Core, Azure IoT Hub, and Google Cloud IoT handle the first three out of the box. Using a managed platform saves 3–6 weeks of backend engineering versus building your own message broker infrastructure. At typical scale, AWS IoT Core costs approximately $1 per million MQTT messages. A 1,000-device fleet sending one reading per minute costs under $50/month in message costs (AWS IoT Core pricing, 2025).

Data processing and application layer

This is where the product differentiation lives: dashboards, alert rules, anomaly detection, and the APIs that integrate with your existing systems. Budget 30–40% of total project cost for this layer. It's often underestimated because clients see it as "just a dashboard" when it's really the product.

Security layer

Security is not optional and can't be retrofitted cheaply. The core security requirements for any production IoT system are device authentication (TLS certificates, not shared secrets), encrypted communications end-to-end, and OTA (over-the-air) firmware update capability with signed updates. Skip these in the build and you'll pay 10–15x to add them later, plus exposure to device compromise at fleet scale (Emproof Industrial IoT Security, 2025). Budget 15–20% of total build cost for security.

The three cost scenarios

Scenario 1: Simple monitoring solution ($20,000–$50,000)

What you get: A single device type, 10–50 units, WiFi or Bluetooth connectivity. Off-shelf hardware with a standard SDK. Cloud backend on a managed platform (AWS IoT Core or Azure IoT Hub). A web dashboard showing live readings, trend charts, and threshold-based alerts. No custom firmware. No mobile app. No third-party integrations.

Who it's for: A small manufacturer wanting to monitor machine temperature or energy consumption. A building manager tracking occupancy across a few floors. A facility team monitoring refrigeration units at a food business. Any situation where you need data visibility but don't yet need complex automation or device management.

Build time: 8–12 weeks

What drives the cost:

• Cloud backend setup and device registry: 2–3 weeks

• Dashboard and alert rules: 3–4 weeks

• Device provisioning and testing: 2–3 weeks

• Security baseline (TLS, device auth): 1 week

Team at RaftLabs rates: 2 engineers x 8–12 weeks = $19,200–$36,000 + project overhead = $20,000–$50,000

Scenario 2: Production IoT system ($50,000–$120,000)

What you get: Multiple device types, 100–10,000 units. Custom firmware or embedded SDK integration. Real-time alerting and anomaly detection. Time-series analytics with trend views and configurable thresholds. Mobile app plus web dashboard. Device management: OTA firmware updates, remote configuration, health monitoring. Third-party integrations (ERP, SCADA, or existing business systems). A proper device provisioning workflow so field teams can onboard new units without engineering involvement.

Who it's for: A logistics company tracking a vehicle fleet. A manufacturer connecting production line sensors to an operational dashboard. A utilities company monitoring distributed infrastructure. A facilities management firm running building automation across 20+ locations. This is where most serious IoT deployments land.

Build time: 14–22 weeks

The 14–22 week range isn't padding. In production IoT projects we've been involved in, 3–5 weeks of that time goes to device provisioning workflows and integration with third-party systems that have inconsistent or undocumented APIs. A project that looks like 14 weeks on paper becomes 18 weeks when your SCADA vendor takes 3 weeks to provision API credentials and their documentation is 4 years out of date.

What drives the cost:

• Custom firmware or SDK integration: 3–5 weeks

• Cloud architecture (time-series DB, stream processing): 2–3 weeks

• Device management and OTA pipeline: 2–3 weeks

• Dashboard, mobile app, and alerts: 4–5 weeks

• Third-party integrations: 2–4 weeks

• Security, testing, and provisioning: 2–3 weeks

Team at RaftLabs rates: 3 engineers x 14–20 weeks = $50,400–$90,000 + PM/QA overhead = $50,000–$120,000

Scenario 3: Enterprise IoT platform ($120,000–$300,000+)

What you get: Large device fleet (10,000+ units). Multi-tenant architecture if you're building a SaaS product on top. Custom firmware with a real-time OS. Edge computing layer for local processing. ML-based anomaly detection and predictive maintenance. Full device lifecycle management (provisioning, decommissioning, firmware rollout by cohort). High-availability architecture with 99.9%+ uptime SLAs. Compliance with relevant standards (IEC 62443 for industrial automation, FDA 21 CFR Part 11 for regulated manufacturing, ETSI EN 303 645 for consumer IoT).

Who it's for: Industrial equipment manufacturers building IoT into their products. Energy companies managing distributed grid assets. Healthcare device manufacturers with regulatory requirements. Any company building an IoT platform as a product, not just using IoT internally.

Build time: 24–40 weeks

Team at RaftLabs rates: 4–5 engineers x 24–36 weeks = $115,200–$225,000 + architect time + compliance overhead = $120,000–$300,000+

What drives the cost up or down

Connectivity choice is a budget decision, not just a technical one

WiFi is the cheapest option to implement and has zero ongoing connectivity cost. Cellular is the most capable but adds $2–$30/device/month in data plan costs. At 500 devices on cellular data plans averaging $10/month, that's $5,000/month in connectivity before you've run a single application query.

The choice matters more at scale. A 50-device WiFi deployment and a 5,000-device cellular deployment have completely different total cost structures, even if the application code looks similar.

Device count changes the testing cost

Per-unit provisioning, testing, and firmware validation is a real engineering cost that scales with device count. Setting up 10 devices is a half-day task. Setting up 500 devices requires a provisioning pipeline, QA processes, and tooling for field technicians. Add 2–4 weeks to the timeline. At 10,000+ devices, you need automated provisioning infrastructure and a device lifecycle management system before you can reliably ship hardware to customers.

Firmware complexity is the biggest single cost driver

Off-shelf hardware with a standard SDK and custom hardware requiring embedded firmware development is a 2–4x cost difference, not a 20% difference. If a device supports the AWS IoT Device SDK or Azure IoT SDK, you write a few hundred lines of code to connect it to the cloud. If you need custom firmware in C/C++ on a microcontroller, you're looking at $30,000–$80,000 in embedded engineering before the cloud and application work even begins. This is the single biggest question to answer when scoping an IoT project.

Data volume changes the infrastructure architecture

100 devices sending one reading per minute is 144,000 messages/day. 10,000 devices sending readings every 10 seconds is 86.4 million messages/day. These require different infrastructure choices, different databases, and different stream processing approaches. Real-time stream processing (Kafka, AWS Kinesis, Azure Event Hubs) adds 20–40% to cloud backend cost compared to batch processing, but enables the sub-second alerting and anomaly detection that makes the system operationally useful.

Security requirements vary significantly by industry

Consumer IoT (smart home, wearables): ETSI EN 303 645 compliance, basic security hygiene, no default passwords, vulnerability disclosure program. Industrial IoT: IEC 62443 defines security levels for operational technology. Healthcare IoT devices used in regulated manufacturing environments: FDA 21 CFR Part 11 requires audit trails, access controls, and validated software. The compliance requirements for industrial and healthcare add 20–35% to total project cost and 4–8 weeks to the timeline.

Ongoing costs

Once your IoT system is in production, the running costs depend on fleet size, connectivity, and how actively you're iterating on the platform.

A 1,000-device fleet on cellular mid-tier plans ($8/device/month) with active cloud usage runs approximately $8,000–$16,000/month in total operating costs. A 100-device WiFi deployment in a fixed location might run under $1,000/month. Size your ongoing cost model before you commit to a connectivity protocol.

Annual maintenance typically runs 15–20% of initial development cost (Guarana Technologies IoT Cost Guide, 2025). For a $80,000 build, budget $12,000–$16,000/year in maintenance and platform evolution.

Build vs. buy: when to use managed IoT platforms

The global IoT market is expected to reach $1.18 trillion by 2026 (Statista IoT Market Forecast, 2025). With that scale comes a mature ecosystem of managed platforms that handle the undifferentiated heavy lifting.

Use a managed platform (AWS IoT Core, Azure IoT Hub, Google Cloud IoT) when:

• You have under 100,000 active devices

• You don't have specialized compliance requirements that the platforms can't satisfy

• You want to move fast. Managed platforms cut 3–6 weeks of backend infrastructure work.

• You're not certain about scale yet. Managed platforms scale elastically without infrastructure engineering.

Build custom infrastructure when:

• You're above 100,000 active devices and the per-message cost of managed platforms is significant

• You have industrial compliance requirements (IEC 62443, FDA 21 CFR Part 11) that require custom audit trail implementations

• You're building a multi-tenant IoT SaaS and the data isolation model of managed platforms doesn't fit your architecture

More than 65% of enterprises are integrating IoT into operational workflows (Fortune Business Insights IoT Report, 2025). The vast majority use managed platforms. Building your own message broker from scratch is almost never the right starting point.

How to get an accurate quote

A real IoT quote requires the vendor to understand:

1. Device type and count: what hardware you're deploying, how many units, and whether it's off-shelf or custom
2. Connectivity protocol: WiFi, cellular, LoRaWAN, or Zigbee, and whether that's already decided or still open
3. Data volume and frequency: how often devices send data and how much of it is time-critical
4. Existing system integrations: what your IoT system needs to read from or write to (ERP, SCADA, existing software)
5. Device management requirements: do you need OTA updates, remote configuration, health monitoring?
6. Security and compliance: what industry you're in and what standards apply

If a vendor quotes you without asking these questions, the number isn't real. The gap between "simple monitoring" and "production fleet management" is $50,000–$100,000 in engineering work. You'll find that gap in scope creep after the project starts.

If you're scoping an IoT build, talk to our team. One call to get a real estimate.

Frequently asked questions

How much does a basic IoT system cost to build?

A simple IoT monitoring solution with 10–50 devices, off-shelf hardware, WiFi or Bluetooth connectivity, and a basic cloud dashboard costs $20,000–$50,000. Build time is 8–12 weeks. The main costs are cloud backend setup, dashboard development, and basic alert rules. Firmware costs are minimal if you're using standard SDKs.

What drives the cost of IoT development?

The five main cost drivers are firmware complexity (off-shelf vs custom embedded), connectivity choice (WiFi vs cellular), device count (per-unit provisioning at scale), data volume and processing requirements (real-time vs batch), and security and compliance needs. Custom firmware alone adds $30,000–$80,000 to a build. It's a 2–4x cost multiplier compared to off-shelf hardware with a standard SDK.

What are the ongoing costs of running an IoT system?

Expect $500–$15,000+/month depending on fleet size and connectivity. Main components are cloud hosting ($200–$5,000/month), cellular data ($2–$30/device/month), maintenance engineering ($2,000–$8,000/month), and monitoring infrastructure ($100–$500/month). A 1,000-device fleet on cellular typically runs $5,000–$12,000/month in total operating costs.

Should I use a managed IoT platform or build custom?

Use a managed platform (AWS IoT Core, Azure IoT Hub) for 90% of builds. They handle device registry, message brokering, security, and OTA updates out of the box and save 3–6 weeks of backend engineering. Build custom infrastructure only above 100,000 active devices when per-message platform costs are material, or if you have compliance requirements the managed platforms can't satisfy.

How long does IoT development take?

A simple monitoring solution takes 8–12 weeks. A production system with device management, custom firmware, and third-party integrations takes 14–22 weeks. An enterprise platform takes 24–40 weeks. The step most teams underestimate is device provisioning and fleet testing. Validating firmware at scale takes 2–4 weeks that can't be shortcut regardless of how simple the individual device logic is.