Market research firm IoT Analytics counted 21.1 billion connected IoT devices in 2025, and it puts cellular at roughly 22% of all IoT connections. That is several billion trackers, meters, cameras, alarms and sensors that depend on a mobile network answering when they ask to connect, and no single mobile network answers everywhere.

Carriers build out where the people are, which leaves gaps along highways, in farm country, at the edge of industrial sites and at every border crossing.

Multi-network IoT data plans are the usual answer to that gap. The phrase gets used loosely in sales material, though, so this piece follows a device through the actual sequence: what happens when it powers on, how it gets permission to use a network, why it does not always pick the strongest signal, what failover really looks like when a network drops, and what changes in countries that do not allow permanent roaming.

First, what a multi-network plan actually is

A multi-network IoT data plan is a cellular subscription on a SIM or eSIM that is allowed to register on more than one mobile network in a given country (and usually across many countries), so the device can move to another network when its current one is weak or unavailable.

A few terms come up repeatedly in the steps below, so it is worth defining them once:

IMSI: the International Mobile Subscriber Identity, a number stored on the SIM that identifies the subscription. Its first digits identify the “home” network that issued it.
Home network and visited network: the home network issued the IMSI; a visited network is any other network the device registers on through a roaming agreement.
Multi-IMSI SIM: a SIM that carries two or more IMSIs from different operators and can switch between them, so it can behave like a local SIM in more than one place.
eUICC: an eSIM chip that can download complete operator profiles over the air, so the operator can be changed after the device ships.
Permanent roaming: a device that lives full time on a visited network rather than visiting for a few days, which is the normal situation for a multi-network IoT SIM and which some countries restrict.

Step 1: Power-on and the first network scan

If you switch on a tracker fitted with a multi-network IoT SIM, the first thing the modem does is read the SIM. The SIM holds the IMSI and a set of lists that tell the modem which networks to prefer. The modem then scans the radio bands it supports and builds a picture of which networks are present and how strong each one is.

The order in which a device chooses is not left to the manufacturer. It is set out in 3GPP TS 23.122, the standard that covers how a device selects a network while it is idle. In automatic mode, the modem works down a ranked list:

The home network (or an equivalent home network), if it is present.
Networks on the user-controlled preference list stored on the SIM.
Networks on the operator-controlled preference list stored on the SIM.
Other networks with a high-quality signal, in random order.
All remaining networks, in order of decreasing signal strength.

This is the first point where the marketing shorthand and the engineering part ways. “Connects to the strongest signal” is a fair summary for a buyer, but in practice the preference lists on the SIM do a lot of the deciding, and signal strength only becomes the main factor further down the list.

Step 2: Getting permission through a roaming agreement

Finding a network is not the same as being allowed onto it. When the device asks to register on a visited network, that network checks the IMSI, sees which home operator issued it, and asks the home operator whether this subscriber is allowed to roam. If there is a roaming agreement in place that covers the device, the answer is yes, and the device attaches.

This is the reason multi-network coverage is really a commercial question as much as a technical one. A multi-network SIM can only use the networks that its issuing operator (or the operator behind the IoT provider) has wholesale agreements with, and those agreements are not easy to get.

Europe’s telecom regulators looked at this directly in the BEREC Report on M2M and permanent roaming (December 2024), which found that access seekers (especially MVNOs) struggle to negotiate permanent roaming agreements, and that quality-of-service terms and SLAs are not widespread in those deals.

If you are comparing plans, the practical takeaway is that “access to hundreds of networks” is only as good as the agreements behind it in the specific countries and regions where your devices will actually sit.

Step 3: Steering, or why the device skips a stronger network

Once a device is roaming, its home operator usually has an opinion about which visited network it should use, normally because some partners are cheaper than others. This is called steering of roaming, and it works in two main ways:

Preference lists on the SIM: the operator-controlled list from Step 1 pushes the device toward partner networks before it considers anything else.
Rejected attach attempts: the home network can refuse registration on a non-preferred network a few times, so the device gives up and tries the next one on the list.

Plans are sometimes sold as “steered” or “non-steered” for this reason. A non-steered plan lets the device settle on whatever network serves it best, which is generally better for coverage and uptime; a steered plan may cost less but can leave a device on a weaker network or make it work through several rejections before it connects. Neither is wrong, but it is a trade-off that buyers should ask about directly, because it changes how the device behaves in the field.

There is a side effect worth knowing. When a network rejects a device in a certain way, the device can record that network on a “forbidden” list kept on the SIM and stop trying it, even if the network would have worked later. A device that is struggling to connect in an area with good coverage on paper is sometimes simply avoiding a network it has been told not to use.

Step 4: The signal drops and failover begins

Failover is the reason most people buy a multi-network IoT SIM, and it is also the step that is most often misunderstood. When the device’s current network fades or goes down, the sequence is roughly:

The modem notices it has lost the network (the radio link fails, or the signal falls below a usable level).
It searches again, using the same ranked order from Step 1, while skipping any networks on the forbidden list.
It registers on the next acceptable network, which means another roaming check with the home operator.
It sets up a new data session, and the device application reconnects to its server.

The important point is that this is rarely instant. Depending on the modem, the radio technology (LTE-M and NB-IoT devices that sleep to save power can take longer to notice a problem), and how many networks it has to try, the gap can run from a few seconds to several minutes. A multi-network plan reduces the chance that a device stays offline; it does not keep a data session alive across the switch.

A second, quieter failure mode catches a lot of teams. Sometimes the radio connection looks healthy but data is not getting through, because of a problem further back in the network. A modem that still sees a strong signal has no reason to move. While the device is roaming, the modem also checks for a higher-priority network on a timer stored on the SIM, which can be set anywhere from minutes to hours, so it will not necessarily move on its own soon.

For device teams, that leads to a short list of things worth doing regardless of which provider is chosen:

Buffer data on the device so that a few minutes offline costs nothing.
Add an application-level check (for example, a heartbeat to your own server) and have the firmware force a network re-scan if the check fails several times in a row.
Do not lock the modem to a single network in firmware, since that defeats the purpose of the plan.
Test failover in the field by actually turning networks off or moving devices into weak areas, rather than trusting the coverage map.

Step 5: When roaming itself is the obstacle

Everything above assumes the device is allowed to roam permanently. In most markets it is. A small number of countries restrict permanent roaming, though, including Brazil and Turkey, and others such as India and China are hard to serve with a roaming identity. A SIM that roams well across North America and Europe may be blocked, or cut off after a period of time, once it is installed in one of those countries.

The BEREC report describes the wholesale options operators use to get around this, including sponsored IMSI roaming and local profiles where permanent M2M roaming is restricted. In device terms, that comes down to three identity models, and each one handles the problem differently:

Single-IMSI roaming: one identity that roams onto partner networks. This is the simplest option, and it works well wherever permanent roaming is allowed.
Multi-IMSI: several identities preloaded on the SIM, with a small application on the SIM switching between them based on location or policy. Switching is fast (minutes), and the SIM can appear as a local subscriber in more places, but the IoT provider decides which IMSIs are on the card.
eUICC: complete operator profiles downloaded over the air. This gives the most flexibility and the best answer for strict markets, but it costs more to integrate and a profile swap takes longer, because a whole profile has to be downloaded. The GSMA’s IoT-specific eSIM standard, SGP.32, had its version 1.2 published in June 2024 and version 1.3 in May 2026, and it is designed for devices with no screen or user that still need their profiles managed remotely.

The approach that tends to hold up across a global rollout is a layered one rather than a choice of one model. The device ships on an eUICC with a multi-network or multi-IMSI profile as its starting (bootstrap) profile, so it connects as soon as it is powered on anywhere in the world, and a local operating profile is pushed over the air only in the markets that need one. Putting this into a single plan is where many deployments stall, because it takes roaming agreements, local profiles and a platform to manage both.

Trafalgar Wireless is one of the providers that has built its multi-network IoT data plans around this layering: its multi-network IoT SIM and eSIM plans give access to more than 500 networks across 180-plus countries through one vendor, at one rate and in one currency per plan, with multi-IMSI and eUICC options available for the countries where permanent roaming is restricted, and all of the SIMs managed from a single platform. For a fleet or logistics business, that means one SKU and one multi-network IoT SIM contract instead of a separate deal with every carrier on the route.

Step 6: Billing, management and security across networks

The last step in the sequence is the one buyers feel every month. Because a multi-network device may spend time on several networks, the way the plan is priced matters as much as the coverage:

Pooled data: the data allowance is shared across all SIMs on the account, so one device that uses more than expected is balanced by others that use less, and overage charges are less likely.
Pay as you use: billing per megabyte plus a fixed monthly fee per SIM, which suits fleets where usage is low or hard to predict.
One rate across networks: some plans charge the same price whichever network or country the device is using; others vary by zone, which makes monthly costs harder to forecast for devices that travel.

Management becomes important as soon as there are more than a few hundred SIMs. A connectivity platform and API should let you activate and suspend SIMs, set data limits and alerts, and see which network each device is attached to, which is often the fastest way to diagnose a failover problem from Step 4.

Security can also carry across networks. In most IoT roaming arrangements, data is routed back through the home operator’s core network before it reaches the internet, so a private APN, a VPN or a static IP address can be applied the same way whichever visited network the device is using at the time. If a provider offers local breakout (where data leaves the network in the visited country to reduce latency), it is worth checking how that interacts with the private APN setup.

Where the sequence matters most

The steps above are the same for every device, but some deployments test them harder than others:

Fleets and trucking: vehicles move through rural stretches and across borders every day, which means frequent rescans and attach attempts.
Emergency and lone-worker devices: a help point or SOS button cannot wait several minutes for a reconnect, so firmware checks and failover testing become essential.
Fixed assets in remote areas: a meter or sensor that never moves still benefits from a second network when the first one has an outage.
Cross-border logistics: containers and cargo trackers are the cases most likely to run into permanent roaming restrictions.

Key takeaways

A multi-network IoT data plan lets one SIM or eSIM register on several networks, but the SIM’s preference lists and the roaming agreements behind it decide which networks a device can actually use.
Network selection follows a ranked order defined in 3GPP TS 23.122; signal strength only becomes the main factor further down that list.
Steering of roaming can keep a device off a stronger network for cost reasons, so ask whether a plan is steered or non-steered.
Failover is a re-scan and re-attach that can take seconds to minutes, so devices need data buffering and an application-level check that can force a re-scan.
Permanent roaming restrictions in countries such as Brazil and Turkey are handled with multi-IMSI or eUICC profiles, and a layered model (a bootstrap profile plus local profiles where needed) covers the most ground.
Pricing model, platform visibility and home-routed security matter as much as the network count once a fleet grows.

How Multi-Network IoT Data Plans Keep Devices Online, Step by Step