Philippine Internet, 2000–2025
Two questions that get answered as if they were one. How many people have a connection — and how fast are the connections that people test? The first comes from the World Bank, the second from 28 quarters of Ookla Speedtest tiles, and this page keeps them apart.
Speeds rose 9.2× in six years. Access did not keep up: the Philippines is 6 of six ASEAN economies for the share of people online.
- The country runs on mobile. 115.3 mobile subscriptions per 100 people against 7.14 fixed broadband lines — about 16 mobile subscriptions for every fixed connection.
- Test-weighted mean fixed download went from 16.25 Mbps in 2019 Q1 to 150.16 Mbps in 2025 Q3. Mobile went from 15.54 to 89.45.
- Fixed latency fell from 52.7 ms to 12.4 ms. Mobile latency stalled at about 28.4 ms and has barely moved since 2021 — the number that decides whether a call or a game feels right, on the connection most people actually use.
- Reported internet use falls from 77.87% in 2023 to 67.26% in 2024. Nobody lost access; the survey changed. The series is not continuous across that break and this page does not draw a smooth line through it.
How Many People Are Online
The World Bank's share of the population using the internet, back to 2000. The line is not continuous: the 2024 figure comes from a different survey basis than 2023 and is more than ten points lower, so the two ends of that step are not comparable.
2024
On the current survey basis.
2023
On the previous basis. The drop between these two is a change in how the question was asked, not people going offline.
2000
Two people in a hundred. Whatever the exact level today, that part of the change is not in doubt.
A Mobile Country, Not A Broadband One
This is the fact that shapes everything else. Fixed broadband barely exists here by regional standards; mobile subscriptions exceed the population. That is not the same as everyone being connected — people hold multiple SIMs — but it does mean the typical connection is a phone on a cell network.
Mobile per 100
More subscriptions than people. Dual-SIM is common, so this overstates the share of people with a phone.
Fixed broadband per 100
About one line for every fourteen people — and a line is usually a household, not a person.
Ratio
Mobile subscriptions per fixed broadband line. Any policy that assumes a wired connection at home is addressing a minority.
Against The Region
The six largest ASEAN economies at 2024, the most recent year every one of them has a figure. Comparing each country at its own latest year would put 2024 against 2019 and read the lag as a gap.
Rank, internet use
Last. Malaysia is at 98.0%.
Rank, fixed broadband
Second from bottom, ahead only of Indonesia.
The gap is access, not speed
On measured speed the Philippines is competitive. On the share of people with a connection it is last of six. Those are different problems and they have different fixes.
Speed, 2019 To Now
Test-weighted mean download from Ookla's Speedtest tiles, two quarters a year. Fixed and mobile are drawn together because the gap between them is the point: they started level and did not stay that way.
Fixed, 2025 Q3
Up from 16.25 Mbps in 2019 Q1 — 9.2×.
Mobile, 2025 Q3
Up from 15.54 Mbps — 5.8×. Real improvement, and now far behind fixed.
They started together
Mbps in 2019 Q1, fixed against mobile. The divergence since is the story of fibre reaching cities while everyone else stayed on a cell tower.
Latency Is The Number That Stalled
Download speed is what gets advertised. Latency — the delay before data starts moving — is what decides whether a video call, a game or a payment terminal feels broken. On fixed lines it improved fourfold. On mobile it has been flat since 2021.
Fixed latency now
Down from 52.7 ms in 2019 Q1.
Mobile latency now
Down from 55.8 ms — but essentially unchanged for four years.
The gap
Mobile latency against fixed. Most people are on the slower side of that ratio, and it is the side that has stopped improving.
North, Central, South
The tiles carry no administrative labels, so grouping them by region would mean inventing one. Latitude bands are what the data actually supports: north of 13°N is Luzon, 10–13°N the Visayas, below 10°N Mindanao.
North (13°N+)
Luzon, including Metro Manila. The fastest band and the most measured.
Central (10–13°N)
The Visayas.
South (<10°N)
Mindanao. The spread across all three bands is smaller than the spread between fixed and mobile — what you connect with matters more than where you are.
What The Speed Numbers Are Not
Ookla's data is the best open measurement of Philippine internet speed there is. It still is not a measurement of the average Philippine connection, and three limits are worth stating before anyone quotes a figure from this page.
These are speeds of tests
People run a speed test when something feels wrong, or right after an upgrade. The sample is self-selected in both directions and is not the population of connections.
A weighted mean, not a median
Every figure here is a test-weighted mean of tile means. Ookla's own headline is a median across tests. The two are different statistics and will not agree; nothing on this page is called "the average speed in the Philippines".
Coverage is where the testing is
A tile exists because someone tested inside it. Mobile tiles fell from 71,732 at their peak to 48,770 in the latest quarter while fixed tiles rose to 120,273. Fewer places measured means the rising mobile line is partly a changing sample, which is why it is reported and not celebrated.
What This Page Does Not Cover
The version this replaces charted internet access by region, device ownership, social media platform shares, online activity mix, adoption by age, e-commerce participation, telco market share, 4G and 5G coverage and a digital literacy rate. None of those came from an open source, and none are reproducible here.
Regional access, age, literacy
PSA collects these in the National ICT Household Survey. Its portal sits behind a managed challenge that scripts do not pass, so nothing here is derived from it.
Social media, devices, online activity
The widely-quoted figures come from DataReportal's annual reports, which are PDF publications built on panel surveys rather than an open dataset. Citable, not reproducible — so they are not charted here.
Telco share and network coverage
Subscriber shares come from operator financial filings and coverage from GSMA, which is licensed. Neither is in this repository.
Method
Two fetchers, seven CSVs, both sources open and keyless.
World Bank
Four indicators from the WDI API: internet use, mobile subscriptions, fixed broadband and fixed telephone. Gaps are left as nulls rather than forward-filled — carrying the last known value to the present asserts a measurement nobody made.
Ookla
Quarterly Speedtest performance tiles on S3, read directly over HTTP by DuckDB. Each tile is about 600 m across and carries the mean speed of tests taken inside it. 28 quarters, filtered to a stated bounding box carried on every row.
A schema change that hid six quarters
Ookla added tile_x/tile_y centroid
columns partway through the archive. Querying them fails on the
older files, so 2019 and 2021 Q3 first came back as
"unavailable" when the data was there. The fetcher now falls back
to parsing the polygon's first vertex, and
checks.sql asserts every year has both connection
types.
Why the two sources stay apart
The World Bank counts people with a connection; Ookla measures connections that were tested. Combining them into one score would be inventing a statistic, so they get separate sections.
Latitude, not region
Speedtest tiles carry no administrative boundary. Bands are used because a region column would have to be made up, the same reasoning the earthquake page uses.
Verification
Fourteen assertions in checks.sql: bounding-box
containment, plausible speeds and latencies, one comparison year
across all six ASEAN countries, no failed quarter, and two
standing warnings for the 2023-24 survey break and the shrinking
mobile sample. Every figure above is bound to a query in
facts.sql.
Key Findings & Summary
- Speed is not the constraint. Test-weighted mean fixed download rose 9.2× in six years, from 16.25 to 150.16 Mbps.
- Access is. The Philippines is 6 of six ASEAN economies for the share of people online and 5 of six for fixed broadband lines per 100 people.
- The typical connection is a phone. 115.3 mobile subscriptions per 100 people against 7.14 fixed lines — roughly 16 to one.
- Mobile latency has been flat at about 28.4 ms since 2021 while fixed fell to 12.4 ms. The measure that decides whether a call works has stopped improving on the connection most people use.
- Geography matters less than connection type. The spread across latitude bands runs from 122.98 to 162.5 Mbps on fixed lines — narrower than the gap between fixed and mobile anywhere in the country.
Sources & Citations
Every figure on this page traces to one of these, through a CSV in
data/ph-internet/. Each is checked against its source query on every
build.
World Bank World Development IndicatorsSecondary
Internet use, mobile and fixed broadband subscriptions, 2000-2024 — Compiled from ITU national reporting
Ookla Open Data (Speedtest tiles)Primary
Fixed and mobile speed and latency by ~600m tile, 28 quarters from 2019 — Aggregated Speedtest results published quarterly as parquet
Let's Discuss This Analysis
Interested in digital connectivity research or ICT policy data analytics?
Have a dataset you'd like analyzed or need a mini AI project as a starter guide? Send me your suggestions!
Prefer the plain-English version?
I wrote a companion post — “Fast Pipes, Not Enough of Them” — in simple, everyday words.