August 2026 is the month SGP.32 stopped being a specification people discussed and became a product people buy. That is the honest headline, and it carries an equally honest asterisk: the SIMs, the orchestration platforms and the first operator services have arrived, but the routers most UK fleets actually run are not SGP.32 routers yet, and will not be for a while. Knowing exactly what has landed, what has not, and when the missing pieces are due is the difference between a connectivity strategy that ages well and one that quietly locks you in for a decade.
The GSMA’s IoT eSIM standard has been the most talked-about development in cellular connectivity for two years running, and for good reason. It finally gives unattended, screenless devices, the meters, trackers, gateways and sensors that make up the bulk of industrial IoT, a proper mechanism for changing operator over the air, from a central platform, with no human at the device. This year is when that promise moved from slideware into commercial delivery. But the market is arriving in pieces, and the pieces are not landing in the order a buyer might hope.
What SGP.32 actually changes
Two earlier standards set the scene. SGP.02 was the original machine-to-machine approach, integration-heavy and prone to lock-in. SGP.22 is the consumer eSIM standard in every modern phone, workable for routers only with a vendor cloud bolted on top, because it assumes a screen and a user to drive it. SGP.32 was written specifically for the headless case. It introduces two new components: the eIM, or eSIM IoT remote Manager, which is the platform an operator or enterprise uses to orchestrate profile changes across a fleet, and the IPA, or IoT Profile Assistant, which executes those instructions on the eUICC itself.
The other quiet but important change is transport. SGP.22 moves profiles over HTTPS and TCP, which is too heavy for an NB-IoT or LTE-M device living on a kilobyte-a-month data budget. SGP.32 specifies a lightweight path, CoAP over UDP with DTLS, that makes remote profile management viable on constrained low-power connections. That single design decision is why SGP.32 matters far more to massive-scale sensor deployments than any consumer eSIM ever could.
Where we are now: the pieces that are live
As of this month the standard is real and certifiable. The stable, certification-grade version of the specification settled in late 2024, and formal GSMA certification is now being issued against it for eUICC chips, eIM platforms and the SM-DP+ servers that store and deliver encrypted profiles. That certification milestone is what unlocked the commercial wave we have seen through the first half of the year.
Networks and operators
The most significant single event is on the operator side. Telenor IoT began commercial delivery of SGP.32 SIM cards in April 2026, making it the first major operator to go fully live rather than merely announce intent. Tele2 IoT is close behind, supporting SGP.32 bootstrap connectivity and running an established hardware partnership with Teltonika. Deutsche Telekom has been active on the connectivity side, and in the automotive world the standard has its first flagship design win, with a major electric-vehicle platform integrating SGP.32 alongside a mobile operator and a security vendor.
The four UK mobile networks, EE, Vodafone, Three and O2, supply the underlying 4G and 5G that all of this runs on, and each is working on a native SGP.32 commercial offering at various stages of readiness. None has a fully productised SGP.32 service in the way Telenor now does, so for a UK enterprise the practical route today runs through the specialist IoT operators and MVNOs rather than a high-street network.
SIMs, chips and platforms
The component layer is genuinely shipping. GSMA-certified SGP.32 eUICC hardware is available from Kigen and Thales. Certified eIM platforms are available from Kigen, Simplex Wireless and others, and the larger connectivity-management players have added SGP.32 orchestration to their existing control planes, with Cisco’s IoT Control Center, emnify’s eIM and Soracom’s multi-operator control layer all now carrying the capability. On the connectivity and reselling side, emnify, KORE, Soracom and Wireless Logic have all moved to put SGP.32 in front of enterprise customers during 2026, several of them announcing end-to-end solutions in the spring.
| Layer | Key players as of Aug 2026 | Status |
|---|---|---|
| Operators live or bootstrapping | Telenor IoT, Tele2 IoT; UK MNOs (EE, Vodafone, Three, O2) building native offerings | Telenor commercially live from April 2026; others staged |
| eUICC hardware (the SIM) | Kigen, Thales, IDEMIA, Giesecke+Devrient | Certified hardware shipping |
| eIM / orchestration | Kigen, Simplex Wireless, Cisco IoT Control Center, emnify, Soracom | Certified platforms available |
| Connectivity / MVNO | emnify, KORE, Soracom, Wireless Logic, Tele2 IoT | Commercial offerings launched through 2026 |
| Router hardware (native SGP.32) | Teltonika, Robustel, Cradlepoint, Digi, Milesight | Mostly SGP.22 today; native SGP.32 on roadmaps |
The router question, answered honestly
Here is where the market gets misread. Walk any major router vendor’s literature today and eSIM is front and centre. That is accurate, but incomplete. The question that decides a procurement is not whether a router has a eUICC chip, it is which specification the management layer implements. As of 2026 the answer almost everywhere is SGP.22 with a proprietary cloud, with native SGP.32 sitting on the roadmap rather than in the box.
The reason is a chain of steps that has to complete before a current eSIM router becomes a genuine SGP.32 router. The modem firmware has to expose the right interface correctly. The router manufacturer has to develop and integrate an in-device IPA, the IPAd. That implementation has to be certified against the SGP.32 test specification. And there have to be commercial agreements with SM-DP+ providers behind it. The first two steps are within the vendor’s control; the rest depend on ecosystem partners, which is exactly why timelines slip.
The practical takeaway for anyone buying kit now: if you purchase the “eSIM variant” of most industrial routers today, you are buying SGP.22 plus a vendor management cloud, not native SGP.32. That is not a bad product, but it is not the standardised, carrier-independent path a long-life deployment may need. Confirm which specification the management layer implements before you commit a fleet.
There is a pragmatic bridge, and it is the most important operational detail in the whole market right now. A removable eUICC in an ordinary plastic SIM form factor, carrying the profile assistant on the card itself as an IPAe implementation, can give a router you already own genuine SGP.32 provisioning with no new hardware and no firmware change. The only requirement is that the modem supports BIP and SIM Toolkit and can run them in automatic mode. Most modern modules can, sometimes after a short configuration tweak over the AT interface. This retrofit route is what lets an estate standardise on SGP.32 across mixed, multi-vendor hardware under a single eIM, rather than waiting years for every vendor to ship native support. The mechanics, the vendor-specific traps and a worked router example are set out in this independent guide to retrofitting a router estate with a plastic eUICC.
On the vendor landscape itself, a few positions are worth noting. Robustel, working with Kigen, is the one manufacturer actively marketing the plastic-retrofit story as a first-class option rather than a workaround. Teltonika, the estate most UK operators actually have on the wall, ships SGP.22 with its RMS platform today and has dipped into SGP.32 via bootstrap connectivity with Tele2, but native support is not yet in the product. Cradlepoint, now under Ericsson, is well placed to move but has not; Digi and Milesight remain SGP.22 for now; and the module-heritage players such as Sierra are evolving. The direction of travel is unanimous. The arrival dates are not.
Modems and modules
Beneath the routers sit the cellular modules, and this is where native SGP.32 will actually be enabled or held back. For an in-device IPAd to work, the module firmware has to expose the eUICC interface properly, and for the plastic-card retrofit route the module has to support BIP and Toolkit in automatic mode. Both are module-level properties, not router features, which is why two units with the same product name but different modem batches can behave differently. The mainstream IoT module vendors, Quectel, Telit Cinterion, Fibocom and Sierra among them, largely have the underlying radio capability; what is still maturing is consistent, certified SGP.32 behaviour across firmware builds. Expect module data sheets to start naming SGP.32 and IPAd support explicitly over the coming year, which is the signal that native router support is genuinely close.
Where we will be in 12 to 18 months
The near-term trajectory is reasonably clear, and it is one of acceleration rather than revolution. Analyst forecasts put the real commercial ramp in the second half of 2026 and through 2027, with one widely cited projection expecting roughly fifty million SGP.32 eSIMs under management globally by 2027, up from a standing start. That is the shape of the next year and a half: not a big-bang switchover, but a steady compounding as certification clears, interoperability improves and more operators follow Telenor into full commercial delivery.
Concretely, over the next 12 to 18 months expect the following. UK network-native SGP.32 offerings should begin to appear as EE, Vodafone, Three and O2 productise what they are currently building, giving domestic enterprises a path that does not route through an overseas operator or a specialist MVNO alone. Cross-vendor interoperability, today’s genuine friction point, should improve as more eUICC, eIM and SM-DP+ combinations complete formal testing against one another, because the failures currently surface at the seams between certified-but-untested component pairings. And the first genuinely native SGP.32 routers and modules should reach the market as manufacturers complete the firmware, IPAd and certification chain, with module data sheets leading the routers by a few months.
The realistic answer on native SGP.32 routers: they are not commercially shipping in volume as of August 2026. The components around them are. Expect the first credible native SGP.32 router hardware to emerge across the next 12 to 18 months, with the plastic-card IPAe retrofit serving as the bridge until then. A deeper vendor-by-vendor breakdown of the hardware and platform landscape is maintained in this independent SGP.32 industrial cellular connectivity guide.
One structural reality should temper any expectation of a clean transition. You cannot upgrade an SGP.22 eUICC to SGP.32 over the air. The eUICC operating system is fixed at manufacture, so moving between standards is a hardware change, not a firmware push. Every embedded-eSIM box already deployed stays on SGP.22 for its working life. Industry guidance is therefore to plan for three to five years of parallel operation, with natural hardware refresh doing the migration rather than any single cut-over. For UK estate owners, that makes the removable-card route more than a stopgap: it decouples the management standard from the hardware-refresh cycle, which is worth real money over a ten-year deployment.
What it means for UK deployers
The summary for anyone planning connectivity in the second half of 2026 is straightforward. The standard is real, certified and commercially delivering at the SIM, platform and first-operator level. The router layer is the laggard, and buying an “eSIM” router today most likely means SGP.22 and a vendor cloud, not the standardised, carrier-independent model SGP.32 promises. If your deployment is long-lived, or if independent eIM control is a procurement requirement, the pragmatic move is to design for SGP.32 now: specify a spare SIM slot, verify modem BIP and Toolkit support, choose a carrier-independent eIM before buying anything, and use plastic IPAe cards to standardise your existing estate while the native hardware catches up over the next year and a half. Get those decisions right at the start, and the flexibility you are paying for survives the decade. Leave them to chance, and you inherit a split estate with several management planes and no easy way back.
Status reflects the market as of August 2026 and is moving quickly; verify current vendor and operator support before any procurement decision. Forecasts referenced are drawn from published analyst commentary (ABI Research, Kaleido Intelligence). This article is independent editorial and names commercial products only for the purpose of describing the market.