Most people who use eSIM every day couldn’t explain how it actually works, and that’s fine — they shouldn’t have to. But if you work anywhere near telecom, it’s worth understanding the plumbing, because the architecture itself is the reason eSIM adoption is accelerating as fast as it is. This isn’t just a smaller SIM card. It’s a fundamentally different model for how a phone talks to a network, and that model is what’s about to expand well beyond phones entirely.
What’s actually inside an eSIM
Here’s the distinction that gets flattened in almost every consumer explainer: “eSIM” describes two related but separate things. One is a physical form factor — a tiny chip (called MFF2) soldered directly onto a device’s circuit board, instead of a removable plastic card. The other, more important one, is software: eUICC, or Embedded Universal Integrated Circuit Card. eUICC is what allows a single physical chip to hold multiple carrier profiles at once and switch between them remotely, without anyone touching the device.
This distinction matters because eUICC is the real innovation. You can technically have an embedded chip that isn’t eUICC-enabled and only holds one permanent profile — that wouldn’t be meaningfully different from a soldered-in traditional SIM. What makes eSIM useful is the software layer that lets it behave like several SIM cards in one.
What happens when you scan that QR code
The activation flow is more interesting than it looks. When you scan an eSIM QR code, you’re triggering a piece of software on your phone called the LPA (Local Profile Assistant), which follows a GSMA specification called SGP.22. The LPA connects to a server on the carrier’s side called the SM-DP+ (Subscription Manager – Data Preparation), authenticates the connection using mutual certificate-based verification — both sides proving who they are before anything is exchanged — and then downloads your carrier profile in encrypted form. The profile is decrypted and installed directly onto the eUICC, and can activate immediately or on a schedule you set.
The security design is worth appreciating: profiles are encrypted end-to-end, and the mutual authentication step means only your specific device, identified by its unique eSIM ID (EID), can decrypt and install a given profile. This is part of why eSIM is generally considered more resistant to certain kinds of fraud than physical SIM swapping, which has been a well-documented attack vector for account takeovers.
Why this architecture enables the growth we’re seeing
The consumer-facing benefits — switch plans without a store visit, hold multiple numbers, activate before you land — are downstream of something more structural: eSIM removes physical SIM cards from the supply chain entirely. Traditional SIM distribution required manufacturing, printing, shipping, warehousing, and retail logistics for a physical object, per carrier, per market. Remote provisioning collapses all of that into a server-side operation. A travel eSIM provider can theoretically serve 190 destinations without holding a single unit of physical inventory anywhere. That’s not a minor efficiency gain — it’s the removal of the single biggest structural cost that kept new entrants out of the connectivity market for two decades.
What comes after eSIM: iSIM
If eSIM was about removing the physical card, the next step — already shipping in some devices — removes the separate chip too. iSIM (Integrated SIM) embeds SIM functionality directly into a tamper-resistant secure enclave within a device’s main system-on-chip, rather than as a distinct soldered component. The functional capability is largely the same as eSIM — remote provisioning, multiple profiles, GSMA-standard security — but the hardware footprint shrinks further, which matters enormously for device categories where every square millimeter is contested: smartwatches, hearables, smart rings, and increasingly, industrial sensors that need to fit inside enclosures with no room for a discrete SIM chip at all.
The IoT and wearables wave nobody’s watching closely enough
Here’s where it gets genuinely interesting for anyone thinking beyond smartphones. Consumer eSIM provisioning (SGP.21/22) assumes a human is present to pull a profile update — you scan the QR code, you approve the install. IoT devices can’t work that way. A shipping container sensor or a connected water meter isn’t going to scan anything. So GSMA built a separate specification, SGP.31/32, which replaces the consumer LPA with an IoT Profile Assistant (IPA) and shifts to a push model — the network operator initiates the profile update remotely, with no human step required at all.
This is the standard quietly becoming foundational infrastructure for an entirely different growth wave. GSMA’s own Mobile Economy 2026 report flags multi-device ownership as one of the defining trends ahead: the same eSIM and 5G principles behind today’s cellular smartwatches are extending to smart rings, health trackers, and body cameras, with AI integration increasingly expected across every one of those touchpoints. On the industrial side, the EU’s eCall mandate — requiring persistent cellular connectivity for automatic emergency calling in new vehicles — is turning eSIM from a nice-to-have into a non-negotiable hardware requirement across the entire European automotive industry. The embedded SIM market, once you include automotive and industrial IoT alongside consumer devices, is projected to grow from $18.7 billion in 2026 to over $73 billion by 2035.
The insider takeaway
The reason eSIM is growing this fast isn’t a single killer feature. It’s that the underlying architecture — remote provisioning, software-defined profiles, standardized security — happens to solve completely different problems for completely different industries at the same time: travel convenience for consumers, fraud resistance for security teams, zero-touch deployment for IoT manufacturers, and space constraints for wearable hardware designers. When one piece of infrastructure quietly solves five unrelated problems at once, that’s usually the signal that it’s not a feature anymore. It’s becoming the default layer everything else gets built on top of.