Travel eSIM Tests Expose Wholesale Routing Latency Gap

Travel eSIM performance can change sharply before a traveller reaches the local mobile network, because wholesale routing decides where the connection exits to the public internet, Ookla's Speedtest analysis of Q2 2026 data shows.

The study mapped more than 50 travel eSIM brands to the wholesale networks behind their SIM profiles, then compared latency and speed against local users in the same destinations.

The result gives a technical view of a market that is often sold through simple retail app brands while the underlying connectivity comes from a smaller set of roaming and wholesale providers.

In the year ending June 2026, Ookla's sample put branded travel eSIM activity on a double-digit growth path, ahead of the 12% rise recorded for total international roaming.

The category's share of measured roaming moved from 2.7% to 3.4%, making routing design a larger performance issue for cross-border mobile data.

A limited set of carriers and roaming specialists sits behind much of the retail market.

Ookla's provider map includes Asian, European and US network-code owners such as Hutchison, KPN, Singtel, Telna, Transatel and Sky, giving the segment a concentrated wholesale base beneath a wider app-store brand layer.

That structure means two customers can buy from different travel eSIM brands, connect to the same visited network and still see very different latency.

Ookla's Speedtest analysis shows Japan's au network at about 54 ms with local breakout and 543 ms when the same radio access used a Madrid-routed path.

Single-hub designs produced the widest delays.

Ookla's analysis shows single-hub penalties of 30 times local latency for Amsterdam-routed profiles in Australia, 25 times for Madrid-routed profiles in Singapore, 18 times in Thailand, 15 times in Japan and 17 times for Singapore-routed profiles in the United States.

Speed caps created a second performance split.

Ookla's distributions placed one large brand around a 6 Mbps download ceiling and one free offer around 0.2 Mbps.

Buyers cannot tell from the retail label alone whether a plan is limited by the visited network, wholesale routing, a commercial cap or a combination of those factors.

Some brands used distributed cores instead of one distant hub.

Saily exited through Germany, the Netherlands, the US, Singapore and Japan, with a global median multi-server latency of 117 ms, while 1GLOBAL used the US and four European hubs and recorded 121 ms.

Ookla's analysis shows Webbing with about 75% of observed traffic using local breakout instead of a remote hub.

Travel eSIM adoption has also become a funded software and distribution market.

Airalo reached unicorn status in July 2025 after a $220 million round and puts its user base above 20 million travellers, while Holafly puts cumulative revenue above $500 million.

Distribution is widening beyond specialist apps.

Apple's eSIM-only iPhone markets have expanded, remote SIM provisioning is becoming routine across device bases, and fintech or travel platforms can add connectivity beside booking flows, foreign exchange or other travel services.

National operators are responding from both sides of the wholesale divide.

T-Mobile sells a prepaid travel eSIM for visitors to the United States, and some operator brands appeared as tenants on Telna's US network code, showing that carrier-branded offers may still rely on third-party travel eSIM infrastructure.

The technical chain starts with the SIM identity.

Ookla used each profile's home-network code to connect the retail app layer with the underlying roaming network.

Earlier academic work had already raised concerns about eSIM egress locations and roaming architecture.

Ookla's contribution is broader measurement from real Speedtest users across many destinations, with the egress map checked against public interconnection records, latency behaviour and server-selection patterns.

Operators face a product-design choice beyond radio coverage.

A cheap single-hub model may be easier to run, while distributed exits or local breakout can change the experience on delay-sensitive services such as maps, messaging, payments, ride-hailing or enterprise authentication.

The measurement leaves the retail market more visible but not fully transparent to buyers.

The public gaps are a universal per-brand ranking table for every destination, plus product-page disclosure on egress locations, wholesale network partners, speed caps and local-breakout availability before purchase.

The remaining buyer risk is contractual rather than radio-based.

A traveller may see the same local carrier name on a phone screen, yet the service path, commercial ceiling and internet exit can still be controlled by an unseen wholesale arrangement.