Summary
- The June 2026 methodology says Ericsson performs traffic measurements in around 100 live networks and calls them "a representative base for calculating worldwide total mobile network traffic", then prints no share of world traffic or subscriptions those networks carry and no factor that converts the sample into the world.
- The same report publishes a subscription series that reconciles exactly with its own printed quarterly additions, which makes it the only quantity a reader can recompute — and the confidence that exact arithmetic earns transfers to the traffic total that has no such check.
The sentence that carries the number is two sentences long, and both are in the method note rather than the headline. "Ericsson regularly performs traffic measurements in around 100 live networks covering all major regions of the world," it reads, and then: "These measurements form a representative base for calculating worldwide total mobile network traffic" (Ericsson Mobility Report, June 2026 — Methodology).
Everything the report is quoted for hangs off that pair. The sample is disclosed — around a hundred networks, spread across the major regions — and the word doing the work is representative, which is an assertion rather than a printed measurement. What the methodology text obtained on 19 November 2026 does not contain is the other half of a scaling statement: no share of the world's subscriptions or of the world's traffic that those hundred networks are said to carry, no weight, and no factor, formula or worked example that takes the sample's measured volume and produces a global figure. A reader can see the sample. A reader cannot see the step from the sample to the world.
The second sentence closes a door that matters more than it looks. "No subscriber data is included in these measurements" (methodology). The subscription figures in the same report are therefore not a by-product of the traffic measurement at all. They are a separate workstream: the methodology describes a subscription and forecast baseline built from historical data from various sources, validated with internal data including customer-network measurements, industry analyst reports, internal assumptions and analyses, and updated subscription figures reported by service providers. No single database is named. No weighting is disclosed. Two of the report's headline families come from two different accounting machines, and only one of them has a stated sample at all.
That leaves four quantities in the June 2026 edition, and they are not the same kind of number.
The traffic point. "Total monthly global mobile network data traffic reached 210 EB," the report says, up 22 percent year on year and about 5 percent quarter on quarter, including fixed wireless access (June 2026 report copy). That copy is third-party hosted; its figures agree with Ericsson's own pages wherever they overlap, and the publisher's addresses could not be read directly on the day.
The table. The key-figures page prints three rows against 2024, 2025 and 2031: 19, 22 and 42 GB per month per smartphone; 123, 146 and 328 EB per month of mobile data traffic; 166, 203 and 515 EB per month of total mobile network traffic including fixed wireless access (Key figures). The rates beside them read 12, 14 and 17 percent. Note that the third row is a different object from the second: one counts mobile data traffic, the other adds fixed wireless access to it. Two definitions, two rows, one report.
The forecast. Mobile data traffic ended 2025 at 146 EB per month and total mobile network traffic at 203, on the way to 328 and 515 in 2031 — roughly 2.2 and 2.5 times — with 5G carrying 48 percent of mobile data traffic at the end of 2025 and 85 percent projected for 2031 (Mobile data traffic outlook). The method note is explicit about what the forecast is: estimated traffic volume in all networks over one month, December (methodology).
The subscription series. 5G subscriptions stood at 2,940 million in 2025 and total mobile subscriptions at 8,840 million, forecast at 6,400 million and 9,540 million in 2031 (June 2026 report copy). Additions are printed too: 162 million net new 5G subscriptions in the first quarter of 2026 took the total to 3.1 billion (press release, 16 June 2026), and 155 million more in the second quarter took it to "close to 3.3 billion" (Mobile subscriptions outlook). The landing page frames the same moment as: "Globally, 5G subscriptions is now close to 3.3 billion, and half of the world's mobile data traffic is carried over 5G" (Mobility Report hub).
The subscription series is the one quantity in this report a stranger can actually recompute. Two thousand nine hundred and forty million, plus 162 million, is 3,102 million, which is what the report prints as 3.1 billion at the first-quarter cutoff; add 155 million and you have 3,257 million, which is what "close to 3.3 billion" means at the second. That is my arithmetic, not the report's reconciliation, and it runs from printed values alone. It proves the series does not contradict itself. It does not vouch for the sources underneath it, which the methodology declines to name individually.
Yet exact arithmetic inside a series is a persuasive thing — it is the reason a reader trusts the next number in the same document, including the one that has no check at all.
The rate column is where the print gets quiet. Twelve, fourteen and seventeen percent sit beside 2024, 2025 and 2031, which invites the reading that each is a six-year growth rate from the first year of the row. It is not. Run the growth from 2025, the middle year: 328 over 146 compounds at 14.4 percent across six years, 515 over 203 at 16.8, and 42 over 22 at 11.4 — the printed 14, 17 and 12. Run it from 2024 and the traffic rows would print 15.0 and 17.5. So the base year of the growth column is the middle of the row, not the left-hand edge, and no sentence in the material obtained says so.
That is my arithmetic, labelled as mine, and it is the only way the printed rates reconcile.
The quarterly point and the annual row are also not the same reference period, and nothing obtained bridges them. Two hundred and ten exabytes is a first-quarter 2026 month; 203 exabytes is a 2025 annual value. The report's own quarter-on-quarter growth of about 5 percent implies a fourth-quarter 2025 figure near 200, which sits comfortably with 203 as a yearly average and tells you nothing about which of the two a chart is plotting. Both are legitimate; neither is the other.
One more check, and it is mine: 146 EB per month divided by 22 GB per month is 6.64 billion. That is a division across two rows of the same table, a count of smartphone-equivalents implied by the table's own ratio. It is not a subscription count, not a measured population, and not something the report claims.
Even the description of the sample turns out to be versioned. A separate, undated forecast-methodology page on the same site puts it differently: "Ericsson regularly performs traffic measurements in over 100 live networks covering all major regions of the world. These measurements form a representative base for calculating world total mobile traffic" (Forecast methodology). Around a hundred and over a hundred are not the same claim, and nothing obtained says which page supersedes the other. That is not evidence that either number is wrong; it is evidence that the sentence describing the instrument is itself an edited object rather than a fixed definition.
And the authorship is a printed fact rather than an insinuation. Ericsson sells the radio access and core network equipment whose demand its forecast describes. That is who writes the number, and it belongs beside the number for the same reason a fund's own valuation belongs beside its net asset value. Nothing read here shows the forecast was shaped to sell anything, and I am not writing that it was.
What is new here is narrow and worth naming. Before reserving this subject I measured the desk's existing coverage on 19 November 2026: a literal search for this report returned no rows, and a thirty-row preview of the thirty-eight articles attached to this company's card contained vendor deals, 5G and 6G announcements, quarterly results, an executive exit and one third-party market-share citation. Not one read the report's sample, its scaling or the status of its quantities. This is a piece about what kind of number each figure is, and about the one arithmetic step the reader is asked to take on trust.
So the practical instruction is not to distrust 210 exabytes. It is to know precisely what you have inherited when you quote it. You have a measurement taken on around a hundred networks that the publisher selects and describes as representative, a global total whose multiplier is not printed, a growth column whose base year sits in the middle of the row, two traffic definitions that differ by fixed wireless access, and a subscription series that adds up because it was built to. The figures are the publisher's instrument. The perimeter is the publisher's too.
Ask for the share and the factor; where they are not printed, quote the number with its provenance attached and stop treating it as a measurement of the world.
Member Briefing
Read the full trend briefing
Members read the full briefing with the source notes behind every finding.
Strategic Circle
A daily briefing in your language, built from changes we have verified.
Join for USD 20/monthCancel anytime.
Members also get the full research library, a daily briefing on the networks they follow and the private forum.






