Human Written & Fact Checked · Updated

186 VoIP Statistics for 2026

VoIP, short for voice over internet protocol, carries phone calls over a broadband connection in place of the old copper telephone network. In the United States, providers reported 62 million interconnected VoIP subscriptions and 14 million traditional switched access lines to the federal regulator for December 2025, and 80.0% of adults lived in homes with mobile phones only in the second half of 2025.

This page gathers VoIP and landline figures from the bodies that count them: US line counts and homes without a landline, copper switch-off plans in the US and the UK, the regulators of France, Germany, Spain, India, Japan and Australia, the UN telecom agency's worldwide estimates, cloud phone providers' own results, calling inside apps, robocalls and caller-ID authentication, telecom fraud, and waits on two tax helplines. Every figure links to the regulator, agency or company that published it, with the date.

The sources measure different things. A count of VoIP subscriptions is not a count of traditional lines, and neither is a share of homes without a landline. A company's seats or revenue is not a regulator's count, a switch-off date is a plan until the network is closed, and calls inside messaging apps are in no regulator's VoIP total. Each country is given on its own regulator's definition and the countries are not ranked.

Headline figures

  • US providers reported 62 million interconnected VoIP subscriptions to the federal regulator for December 2025, a count taken from mandatory provider filings. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • The US had 14 million end-user switched access lines, the regulator's term for traditional phone lines, in service in December 2025. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • 68% of the 75 million US wireline retail voice connections (switched access lines plus interconnected VoIP) in December 2025 were business connections, 51 million in all. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • 80.0% of US adults lived in wireless-only households, with no landline, in July-December 2025, on the federal health interview survey. (CDC (NHIS wireless substitution, Jul-Dec 2025), Sep 2026)
  • The UK government says over two thirds of UK landlines have already been upgraded to VoIP, with all landlines needing to be upgraded by January 2027 under an industry plan. (UK Government (Moving landlines to digital technologies), May 2025)
  • AT&T said in October 2026 that more than 100 of its US wire centres had no customers left on legacy services; it targets ending copper-based services in 50% of its wire centres outside California by the end of 2027 and substantially all of them by the end of 2029. (AT&T (Q3 2026 earnings call), Oct 2026)
  • Spain completed the switch-off of its copper access network in May 2025, a process begun in 2015; the Spanish regulator says Spain is the first European Union country to do so. (CNMC (annual sector report 2025), Jun 2026)
  • France had 35.7 million fixed telephone subscriptions at 30 June 2026, of which 94% were voice over broadband, the French regulator's term for VoIP lines. (Arcep (electronic communications observatory, Q2 2026), Oct 2026)
  • 37.86 million of Germany's 37.9 million fixed-network voice accesses were IP-based at the end of 2025, and the German regulator calls the switch from analogue and ISDN lines to VoIP "almost complete". (Bundesnetzagentur (annual telecoms report 2025), Jun 2026)
  • Worldwide fixed-telephone subscriptions have shrunk by 3% a year over the past five years, on the UN telecom agency's estimates for 2025. (ITU (Facts and Figures 2025), Nov 2025)
  • US consumers received an estimated 52.5 billion robocalls in 2025, less than 1% below the 52.8 billion of 2024, on a private index extrapolated from one call-blocking app's users. (YouMail (Robocall Index, 2025 full year), Jan 2026)
  • The US Federal Trade Commission received 2,618,077 Do Not Call complaints in fiscal year 2025 (October 2024 to September 2025). (FTC (Do Not Call Data Book FY 2025), Dec 2025)
  • 53.9% of calls were signed with STIR/SHAKEN caller-ID authentication when they reached the terminating provider in October 2026, among US calls received by customers of the vendor TransNexus. (TransNexus (STIR/SHAKEN statistics, October 2026), Nov 2026)
  • Telecoms fraud cost the industry an estimated $41.82 billion worldwide, 2.46% of revenues, in the 2025 edition of an industry body's survey of operators. (CFCA (Global Fraud Loss Survey 2025), Mar 2026)

How many VoIP lines are there in the United States?

Tens of millions, though the count has been slipping for three years. Interconnected VoIP is now the main kind of fixed phone line in the regulator's provider filings, well ahead of traditional switched lines.

66,241 Dec 2022 65,675 Jun 2023 64,437 Dec 2023 64,651 Jun 2024 63,876 Dec 2024 63,397 Jun 2025 61,778 Dec 2025
View data table
US interconnected VoIP subscriptions at each half-year from December 2022 to December 2025, in thousands, as printed in the FCC's December 2025 edition.
PeriodValue
Dec 202266,241
Jun 202365,675
Dec 202364,437
Jun 202464,651
Dec 202463,876
Jun 202563,397
Dec 202561,778
US interconnected VoIP subscriptions at each half-year from December 2022 to December 2025, in thousands, as printed in the FCC's December 2025 edition. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • Mobile subscriptions stood at 397 million in December 2025, and switched lines, interconnected VoIP and mobile together made 472 million retail voice telephone service connections in the United States. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • Over the three years to December 2025, US interconnected VoIP subscriptions declined at a compound annual rate of 2.3%, on the FCC's own calculation. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • US mobile voice subscriptions increased at a compound annual rate of 2.1% over the same three years. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • Over-the-top interconnected VoIP, where the customer's broadband comes from a different company, accounted for 28,654 thousand of the 61,778 thousand US interconnected VoIP subscriptions in December 2025. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • Incumbent local exchange carriers held 5,320 thousand US interconnected VoIP subscriptions in December 2025 and non-incumbent providers 56,458 thousand. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • 2,415 providers reported US wireline voice subscriptions, switched access lines or interconnected VoIP, to the FCC for December 2025, of which 678 were incumbent local exchange carriers. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • In June 2017 the same report series counted 64 million interconnected VoIP subscriptions, 55 million end-user switched access lines and 336 million mobile subscriptions in the US. (FCC (Voice Telephone Services, June 2017), Nov 2018)
  • An earlier edition of the report printed 66,531 thousand US interconnected VoIP subscriptions for June 2019 and 68,035 thousand for June 2022. (FCC (Voice Telephone Services, June 2022), Aug 2023)

These are counts from FCC Form 477, the filing every US voice provider must make. Interconnected VoIP is the FCC's regulatory term for a service that can call and be called from the public telephone network, so calls that stay inside an app are not in it. Each edition revises earlier periods: the chart uses the December 2025 edition, and the 2017 and 2019-2022 figures come from older editions that may since have been revised.

How many traditional landlines are left in the United States?

Few, and fewer every half-year. Traditional switched access lines are falling far faster than VoIP subscriptions in the regulator's count, and most of those left still run over copper.

24,412 Dec 2022 22,004 Jun 2023 20,680 Dec 2023 18,234 Jun 2024 16,413 Dec 2024 15,025 Jun 2025 13,528 Dec 2025
View data table
US retail switched access lines at each half-year from December 2022 to December 2025, in thousands, as printed in the FCC's December 2025 edition.
PeriodValue
Dec 202224,412
Jun 202322,004
Dec 202320,680
Jun 202418,234
Dec 202416,413
Jun 202515,025
Dec 202513,528
US retail switched access lines at each half-year from December 2022 to December 2025, in thousands, as printed in the FCC's December 2025 edition. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)

A switched access line is defined by how the call is switched, not by the wire, so some run over fibre. The 2019-2022 figures come from an older edition of the report and may since have been revised.

When will AT&T switch off its copper network?

  • AT&T said in December 2024 that only 5% of its residential customers still used copper voice technology. (AT&T (2024 investor day), Dec 2024)
  • At the same event AT&T said it expected to no longer provide copper-based services across the vast majority of its footprint by the end of 2029, a company plan. (AT&T (2024 investor day), Dec 2024)
  • AT&T's copper network comprised 4,600 wire centres across a wireline footprint of over 500,000 square miles as of December 2024. (AT&T (2024 investor day), Dec 2024)
  • By December 2024 AT&T said it had stopped offering regulated legacy services across about 250,000 square miles. (AT&T (2024 investor day), Dec 2024)
  • AT&T plans to have no copper customers by the end of 2027 in what it calls its Wireless First wire-centre areas, about 50% of its land area and 10% of the population. (AT&T (2024 investor day), Dec 2024)
  • AT&T said in July 2026 that the FCC had given it permission to discontinue legacy copper voice service in about 60% of its wire centres in California. (AT&T (Q2 2026 prepared remarks), Jul 2026)
  • In the same remarks AT&T said it had approval to discontinue legacy services in over 30% of its wire centres nationwide, effective by late 2026. (AT&T (Q2 2026 prepared remarks), Jul 2026)
  • Service revenues in AT&T's Legacy segment, its copper-based voice and data services, fell about 25% year over year in the third quarter of 2026, and the company expects continued declines of more than 20% as it powers down copper. (AT&T (Q3 2026 earnings call), Oct 2026)

These are AT&T's own statements to investors. The 2027 and 2029 dates are targets, and the shares are of wire centres (local exchanges), not of customers or lines. The FCC figures above are the regulator's count for all US providers.

Is a VoIP line more likely to be in a home or a business?

A business. Most US wireline voice connections are business lines, and over-the-top VoIP, the kind sold by cloud phone providers, is almost entirely a business product.

  • Coaxial cable 16,441
  • Copper local loop 8,716
  • Fibre to the premises 7,827
  • Fixed wireless or satellite 141
US interconnected VoIP subscriptions that are not over-the-top, by the last-mile medium that delivers them, December 2025, in thousands. Cable carries about half; over-the-top VoIP is left out because no medium is reported for it. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • Residential connections were 24 million, or 32%, of US wireline retail voice connections in December 2025. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • US business customers held 43,515 thousand interconnected VoIP subscriptions and 7,733 thousand switched access lines in December 2025, a total of 51,248 thousand. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • US residential customers held 18,263 thousand interconnected VoIP subscriptions and 5,795 thousand switched access lines in December 2025, a total of 24,058 thousand. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • Of the over-the-top interconnected VoIP subscriptions in the US in December 2025, 27,219 thousand were business and 1,435 thousand residential. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • Interconnected VoIP that is not over-the-top, which the FCC lists as all other VoIP, totalled 33,124 thousand US subscriptions in December 2025, of which 16,441 thousand ran over coaxial cable. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • The US had 0.18 residential wireline voice connections per household in December 2025, counting switched access lines and interconnected VoIP together. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • Among the states in the FCC's table, Vermont had 0.35 residential wireline connections per household in December 2025, New Jersey 0.31, and Massachusetts, New York and Delaware 0.30 each. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • Connecticut and New Hampshire each had 0.29 residential wireline connections per household in December 2025 and West Virginia 0.28. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • California had 0.19 residential wireline connections per household in December 2025, Florida 0.15, Illinois 0.14 and Arizona 0.12. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)
  • Idaho, Utah and Arkansas each had 0.11 residential wireline connections per household in December 2025, and Texas, Oklahoma and Mississippi 0.10. (FCC (Voice Telephone Services, Dec 2025), Nov 2026)

The FCC's business category covers business and government grade service. Over-the-top means the customer's broadband comes from a company not affiliated with the VoIP provider. The per-household ratios divide residential connections by household counts from the census survey; the states named are a selection from the FCC's table and are not ranked.

How many US households have no landline?

Most of them. Four in five US adults live in a home with mobile phones only, on the federal health survey that tracks it. The survey counts a VoIP home phone as a landline, so it measures homes with no fixed phone of any kind, as a share of people and not a count of lines.

76.0% Jul-Dec 2023 77.1% Jan-Jun 2024 78.7% Jul-Dec 2024 78.9% Jan-Jun 2025 80.0% Jul-Dec 2025
View data table
Share of US adults living in wireless-only households, by half-year, from the federal health interview survey. Changes between neighbouring half-years can fall inside the survey's confidence intervals.
PeriodValue
Jul-Dec 202376.0%
Jan-Jun 202477.1%
Jul-Dec 202478.7%
Jan-Jun 202578.9%
Jul-Dec 202580.0%
Share of US adults living in wireless-only households, by half-year, from the federal health interview survey. Changes between neighbouring half-years can fall inside the survey's confidence intervals. (CDC (NHIS wireless substitution, Jul-Dec 2025), Sep 2026)

Early release estimates from the National Health Interview Survey of civilian noninstitutionalized US adults (9,865 adults in July-December 2025). A landline here is any home phone that is not a mobile, VoIP included. Household status (what phones the home has) and personal status (how the adult uses them) are separate tables, so the two all-adult shares differ. Shares of people, not counts of lines.

When will UK landlines be switched off?

By January 2027 on the industry's plan, which is a target and not yet a completed closure. This page has no regulator line count for the UK, so the counts below are the incumbent BT's own and cover only its business unit and its access network.

869 Sep 2024 711 Mar 2025 538 Sep 2025 371 Mar 2026 314 Sep 2026
View data table
Traditional analogue voice lines at BT's Business unit at each half-year end, in thousands, on the company's own count. This is one unit of one UK provider, not the UK total.
PeriodValue
Sep 2024869
Mar 2025711
Sep 2025538
Mar 2026371
Sep 2026314
Traditional analogue voice lines at BT's Business unit at each half-year end, in thousands, on the company's own count. This is one unit of one UK provider, not the UK total. (BT Group (H1 FY27 KPIs), Nov 2026)
  • The UK's old phone network, the PSTN, had over 2,600 major incidents in 2024/25, the government says. (UK Government (Moving landlines to digital technologies), May 2025)
  • The same government page, citing the telecoms regulator, says 2024 saw a 45% increase in the number of PSTN incidents reported. (UK Government (Moving landlines to digital technologies), May 2025)
  • UK guidance issued in 2018 says phone providers should have at least one solution that lets customers reach emergency services for a minimum of one hour during a power cut at the premises, offered free to people who depend on their landline. (UK Government (analogue to digital landlines guidance), Feb 2024)
  • BT Group's Business unit had 1,699 thousand voice lines at 30 September 2026: 314 thousand traditional analogue voice lines and 1,385 thousand VoIP seats. (BT Group (H1 FY27 KPIs), Nov 2026)
  • BT Business had 869 thousand traditional voice lines in September 2024, 711 thousand in March 2025, 538 thousand in September 2025 and 371 thousand in March 2026. (BT Group (H1 FY27 KPIs), Nov 2026)
  • VoIP seats at BT Business numbered 1,336 thousand in September 2024, 1,374 thousand in March 2025, 1,381 thousand in September 2025 and 1,391 thousand in March 2026. (BT Group (H1 FY27 KPIs), Nov 2026)
  • On a wider measure, units on legacy networks at BT Business fell 53% year on year, from 821 thousand to 384 thousand, in the half year to 30 September 2026. (BT Group (H1 FY27 results), Nov 2026)
  • BT Group expects a headwind of about £150-200 million to adjusted service revenue from voice in its financial year to March 2027 as it moves towards PSTN switchover, a company forecast. (BT Group (H1 FY27 results), Nov 2026)
  • Openreach, BT Group's UK access network arm, had 19,927 thousand physical lines at 30 September 2026, down from 22,315 thousand at 30 September 2024. (BT Group (H1 FY27 KPIs), Nov 2026)
  • Openreach's broadband lines at 30 September 2026 were 9,901 thousand full fibre (FTTP), 8,051 thousand VDSL and 956 thousand ADSL, a total of 18,908 thousand. (BT Group (H1 FY27 KPIs), Nov 2026)

The government's statement on upgraded landlines carries no date or count, and its incident figures are relayed from the regulator. BT's figures are company statements for BT Business and Openreach only; other UK networks are not included. BT's units on legacy networks is a wider measure than its traditional voice lines, so the two are not the same count. ADSL and VDSL run wholly or partly over copper.

How many fixed lines in France run over VoIP?

Nearly all of them. The French regulator counts voice over broadband separately from the old switched network, and the switched network is shrinking fast as copper lines are replaced.

2.939m Q2 2025 2.770m Q3 2025 2.579m Q4 2025 2.395m Q1 2026 2.248m Q2 2026
View data table
Fixed telephone subscriptions on France's traditional switched network at the end of each quarter, in millions. Voice-over-broadband subscriptions are a separate count and are not in this chart.
PeriodValue
Q2 20252.939m
Q3 20252.770m
Q4 20252.579m
Q1 20262.395m
Q2 20262.248m
Fixed telephone subscriptions on France's traditional switched network at the end of each quarter, in millions. Voice-over-broadband subscriptions are a separate count and are not in this chart. (Arcep (electronic communications observatory, Q2 2026), Oct 2026)

Figures come from the French regulator's quarterly survey of operators and are provisional until the next edition. Voice over broadband is the regulator's term for a phone service on a fibre, cable, DSL or fixed 4G or 5G access. Lines and subscriptions are separate counts: one line can carry more than one subscription. The never-use and daily-mobile shares come from a survey of people, not from operators.

Has Germany finished moving its landlines to VoIP?

Almost, on the regulator's own description. What keeps changing is how little those lines are used: call minutes on German fixed networks fell sharply in 2025.

  • Within national fixed networks about 39bn
  • To national mobile networks about 7bn
  • Abroad about 1bn
Where Germany's fixed-network call minutes went in 2025, in billions of minutes, on the regulator's rounded figures. Most calls from a fixed line still end on another fixed line. (Bundesnetzagentur (annual telecoms report 2025), Jun 2026)

Year-end counts from operators' returns to the German regulator. Call-by-call and preselection services ended on 31 December 2024, which affects how 2025 fixed minutes compare with earlier years. Fixed and mobile minutes here are calls on numbered phone services; calls inside messenger apps are counted separately and appear in the app-calling section.

How much calling in Germany is on mobile networks?

The VoLTE figure counts users able to make calls over the 4G network, not the minutes carried that way.

How many fixed lines does Spain have now its copper network is closed?

Spain's fixed lines are falling slowly, and none of them run on copper any more. The closure is complete, not planned, so every remaining line is carried over fibre, cable or another access.

  • Residential 12.8m (70%)
  • Business 5.4m (30%)
Spain's fixed telephone lines at the end of 2025 by customer type, in millions, on the regulator's annual report. Seven lines in ten are residential. (CNMC (annual sector report 2025), Jun 2026)

Line counts come from the Spanish regulator's operator data; monthly 2026 figures are provisional until the next annual report. The never-use share comes from its household panel, a survey. The statement that Spain was the first EU country to close its copper network is the regulator's own. The report gives two different values for the share of households with fixed telephony, so that figure is left out.

Are fixed telephone lines disappearing worldwide?

They are shrinking on the UN telecom agency's estimates, while fixed broadband keeps growing. The agency counts a VoIP line with a fixed number as a fixed-telephone subscription, so the fall covers every kind of fixed line, copper or not. India and Japan follow, each on its own regulator's definition.

47.49m Jun 2025 46.61m Sep 2025 47.37m Dec 2025 48.25m Mar 2026 47.82m Jun 2026
View data table
Wireline telephone subscribers in India at the end of each quarter, in millions, on the Indian regulator's quarterly report. The regulator's later monthly release, covered below, runs to the end of September 2026.
PeriodValue
Jun 202547.49m
Sep 202546.61m
Dec 202547.37m
Mar 202648.25m
Jun 202647.82m
Wireline telephone subscribers in India at the end of each quarter, in millions, on the Indian regulator's quarterly report. The regulator's later monthly release, covered below, runs to the end of September 2026. (TRAI (performance indicators, April-June 2026), Oct 2026)
  • Worldwide fixed broadband subscriptions grew by an average of 5.2% a year over the past five years, on the agency's 2025 estimates. (ITU (Facts and Figures 2025), Nov 2025)
  • Fixed broadband, at 20 subscriptions per 100 inhabitants worldwide in 2025, has passed the historical peak of 19 fixed-line telephone subscriptions per 100 inhabitants recorded in 2006. (ITU (Facts and Figures 2025), Nov 2025)

The worldwide rates are the UN telecom agency's estimates. India's wireline count covers fixed telephone connections of every technology, from operators' monthly returns. Japan's fixed telephone count includes IP telephones with ordinary geographic numbers, which are counted by numbers in use. The definitions differ, so the countries are not compared with each other.

How many wireline phone subscribers does India have?

The monthly release is newer than the quarterly report, and both are correct for their dates. Internet telephony minutes cover licensed providers reporting to the regulator only, not calls inside messaging apps. Providers are listed as printed, not ranked.

How many fixed and IP phone lines does Japan have?

Ministry statistics from operators' statutory reports. IP telephones are counted by numbers in use, not by households or lines. 050 numbers are non-geographic IP telephone numbers.

How many Australians still use a landline?

A small and shrinking minority, with older people far more likely to. This is the communications regulator's yearly survey of adults, so it measures who used a home landline for calls and not how many lines exist.

54.2% 2017 48.4% 2018 44.0% 2019 40.3% 2020 24.4% 2021 23.0% 2022 18.3% 2023 14.9% 2024 12.1% 2025
View data table
Share of Australian adults who used a home landline for personal calls in the previous 6 months, by survey year, from the regulator's data tables (rounded to one decimal). The regulator enlarged the survey sample from 2021 and reworded the question over the years, so read the step between 2020 and 2021 with care.
PeriodValue
201754.2%
201848.4%
201944.0%
202040.3%
202124.4%
202223.0%
202318.3%
202414.9%
202512.1%
Share of Australian adults who used a home landline for personal calls in the previous 6 months, by survey year, from the regulator's data tables (rounded to one decimal). The regulator enlarged the survey sample from 2021 and reworded the question over the years, so read the step between 2020 and 2021 with care. (ACMA (How we communicate, data tables), Feb 2026)

A consumer survey of Australian adults about personal use in the 6 months before each yearly survey. Use of a landline is not the same as having one, and the survey does not count lines.

How big are the cloud phone providers?

Big enough to report billions of dollars of revenue and millions of seats, but each company counts in its own way. These are company statements for their own fiscal periods, not regulator counts, and they are not ranked against each other here.

1m 2021 (licenses) 4m 2022 (seats) 10m 2025 (seats)
View data table
Zoom Phone milestones as listed on Zoom's company timeline, in millions, by the year each was reached. Zoom uses licenses for 2021 and seats after; these are announcements, not year-end counts. The product launched in 2018.
PeriodValue
2021 (licenses)1m
2022 (seats)4m
2025 (seats)10m
Zoom Phone milestones as listed on Zoom's company timeline, in millions, by the year each was reached. Zoom uses licenses for 2021 and seats after; these are announcements, not year-end counts. The product launched in 2018. (Zoom (company timeline), 2026)
  • RingCentral reported total revenue of $676 million for the third quarter of 2026 (quarter ended 30 September 2026), up from $639 million a year earlier, growth of 5.9%. (RingCentral (Q3 2026 results), Oct 2026)
  • RingCentral's subscriptions revenue was $653 million in the third quarter of 2026, up 6.0% and 97% of total revenue. (RingCentral (Q3 2026 results), Oct 2026)
  • RingCentral said annual recurring revenue from customers using one of its native paid AI products was approaching 15% of the total in the third quarter of 2026, doubling year over year, on its own metric. (RingCentral (Q3 2026 results), Oct 2026)
  • RingCentral's guidance for full-year 2026, a company forecast issued in October 2026, is total revenue of $2.650 billion to $2.657 billion and subscriptions revenue of $2.563 billion to $2.569 billion. (RingCentral (Q3 2026 results), Oct 2026)
  • Zoom said on 9 October 2025 that Zoom Phone, its cloud phone system, had recently surpassed 10 million seats globally. (Zoom (Zoom Phone 10 million seats release), Oct 2025)
  • Zoom said in September 2022 that it had surpassed 4 million Zoom Phone seats sold the month before, 3.5 years after launch, and had sold more than 700,000 licenses in the last quarter alone. (Zoom (Zoom Phone blog, 2022), Sep 2022)
  • Microsoft said Teams Phone reached 12 million PSTN users worldwide in June 2022, nearly double the number a year earlier; these are users with a phone number assigned through one of its calling options. (Microsoft (Teams Phone blog, 2022), Aug 2022)
  • Bandwidth, a US cloud communications provider, reported revenue of $234 million in the third quarter of 2026, up 22% year over year, which it called a record, and guided to full-year 2026 revenue of $905-910 million, a forecast. (Bandwidth (Q3 2026 results), Oct 2026)
  • Twilio, a US communications-API provider covering voice, messaging and email, reported revenue of $1.56 billion in the third quarter of 2026, up 20% year over year (16% organic). (Twilio (Q3 2026 results), Oct 2026)
  • Twilio's dollar-based net expansion rate was 114% in the third quarter of 2026, against 109% a year earlier. (Twilio (Q3 2026 results), Oct 2026)
  • Five9, a US cloud contact-centre software provider, reported revenue of $323.3 million in the third quarter of 2026, up 13% from $285.8 million a year earlier, with subscription revenue up 16%. (Five9 (Q3 2026 results), Nov 2026)
  • Five9 expects full-year 2026 revenue of $1.269 billion to $1.275 billion, a company forecast. (Five9 (Q3 2026 results), Nov 2026)

All figures are the companies' own, mostly from unaudited quarterly results. Revenue covers every product each company sells: Twilio does not split out voice, and Bandwidth's total includes messaging surcharges it passes through. Seats, licenses and PSTN users are different units defined by each company. Guidance figures are forecasts. Microsoft's figure dates from 2022.

How much calling happens inside apps?

A great deal, and none of it shows up in regulators' VoIP subscription counts. Calls inside messaging apps are over-the-top services that need no phone number, so they are reported separately where they are reported at all. In Germany, where the regulator does collect them, messenger voice minutes exceed fixed and mobile call minutes.

  1. 1 Voice calls 198bn
  2. 2 Video calls 188.5bn
Minutes of outgoing voice and video calls over messenger and video-conference services in Germany in 2025, in billions, as collected by the German regulator from providers. (Bundesnetzagentur (annual report release), Jun 2026)

WhatsApp's two figures are not like for like: the 2016 figure counts voice calls and the 2024 figure counts voice and video calls. The German figures cover services that do not use phone numbers, collected by the regulator from providers. The Australian shares come from a survey of adults. None of this is counted in the FCC's interconnected VoIP subscriptions.

How many robocalls do Americans receive?

Billions every month on one private index, though no regulator counts them. The index is an estimate built from one call-blocking app's users and includes wanted automated calls; complaints to the consumer regulator are a separate count of reports filed, and the two cannot be compared as levels.

  • Telemarketing and scams 1.75bn (44%)
  • Notifications 1.65bn (41%)
  • Payment reminders 0.69bn (15%)
Estimated US robocalls in October 2026 by type, in billions, on the index publisher's own classification. Notifications and payment reminders are automated calls people may want. (YouMail (Robocall Index, October 2026), Nov 2026)

The robocall volumes are one firm's estimate, extrapolated from robocall traffic to users of its call-blocking app, not a network count. Do Not Call complaints are unverified reports from consumers. One is calls placed and the other is reports filed.

How many people complain about unwanted calls?

  • Of the Do Not Call complaints the FTC received in fiscal year 2025, 1,601,611 were about robocalls, 802,144 about live callers and 214,322 did not report the call type. (FTC (Do Not Call Data Book FY 2025), Dec 2025)
  • The US National Do Not Call Registry held about 258.5 million active registrations on 30 September 2025, after more than 4.7 million phone numbers were added during the fiscal year. (FTC (Do Not Call data book release), Dec 2025)
  • Do Not Call complaints to the FTC in fiscal year 2025 remained about 48% lower than in fiscal year 2021, when the FTC received approximately five million reports. (FTC (Do Not Call data book release), Dec 2025)
  • The FTC's top five states for Do Not Call complaints per 100,000 people in fiscal year 2025 were Arizona (1,028), Tennessee (1,017), Nevada (960), Illinois (943) and Florida (933). (FTC (Do Not Call data book release), Dec 2025)
  • Since the National Do Not Call Registry was set up in 2003, the FTC has filed 173 lawsuits against 570 companies and 449 individuals over unwanted telemarketing calls, collecting nearly $400 million. (FTC (biennial Do Not Call report release), Jan 2026)

Registrations are phone numbers, not people. The state ranking is the FTC's and counts complaints, not calls. The $400 million is money collected from 2003 to January 2026, not judgments ordered.

Does caller-ID authentication stop spoofed calls?

Not on its own. A signature shows which provider vouched for a call, not that the call is honest, and on one vendor's traffic a large share of calls still arrive unsigned. Regulators also police which providers may carry calls at all.

  1. 1 Signed, A-level 1.8%
  2. 2 Signed, B-level 4.6%
  3. 3 Signed, C-level 8.9%
  4. 4 All calls 4.0%
Share of calls that were robocalls in October 2026, by STIR/SHAKEN attestation level, on TransNexus customers' networks and the vendor's own robocall classification. A signed call is not always a legitimate one. (TransNexus (STIR/SHAKEN statistics, October 2026), Nov 2026)
  • In October 2026, 31.9% of calls received by TransNexus customers carried full A-level STIR/SHAKEN attestation, 5.7% B-level and 9.8% C-level. (TransNexus (STIR/SHAKEN statistics, October 2026), Nov 2026)
  • Robocalls made up 4.0% of all calls on TransNexus customers' networks in October 2026, on the vendor's own classification. (TransNexus (STIR/SHAKEN statistics, October 2026), Nov 2026)
  • 10,872 voice providers had filings in the FCC's Robocall Mitigation Database as of 21 April 2026. (FCC (FCC 26-32 rulemaking notice), May 2026)
  • The FCC found that Lingo Telecom applied A-level attestations to 3,978 spoofed robocalls carrying a deepfake AI voice message purporting to be from then-President Biden. (FCC (FCC 26-32 rulemaking notice), May 2026)
  • The FCC proposed a $4,500,000 penalty against Voxbeam Telecommunications in April 2026 for accepting calls from a foreign provider not listed in the Robocall Mitigation Database; the penalty is proposed, not final. (FCC (Voxbeam notice of apparent liability), Apr 2026)
  • The FCC's notice says Voxbeam transmitted 60,873 calls to US numbers from over 2,812 distinct US caller IDs over a few days from 31 March 2025, many spoofing numbers of large US banks. (FCC (Voxbeam notice of apparent liability), Apr 2026)

The signed-call and attestation shares are one vendor's measured traffic: calls received by the US voice providers that use its software, not all US calls. The three attestation shares are quoted as printed and do not sum to the signed share. The Voxbeam penalty is a notice of apparent liability, a proposal the company can contest.

How much does telecoms fraud cost?

  • The 2025 estimate of worldwide telecoms fraud losses was up $2.87 billion from $38.95 billion in 2023. (CFCA (Global Fraud Loss Survey 2025), Mar 2026)
  • The top two fraud methods in the 2025 survey were subscription fraud with true or stolen identities ($5.31 billion) and first-party subscription fraud ($4.89 billion). (CFCA (Global Fraud Loss Survey 2025), Mar 2026)
  • The top two fraud types in the 2025 survey were device and hardware reselling ($4.18 billion) and spoofing or caller-ID manipulation ($3.55 billion, a new entry). (CFCA (Global Fraud Loss Survey 2025), Mar 2026)
  • 76% of respondents to the 2025 telecoms fraud loss survey said fraud losses globally had trended up over the past 12 months, against 53% in 2023. (CFCA (Global Fraud Loss Survey 2025), Mar 2026)
  • For their own company, 35% of respondents to the 2025 survey said fraud losses had trended up and 38% said they had trended down, against 9% saying down in 2023. (CFCA (Global Fraud Loss Survey 2025), Mar 2026)

The loss figure is an estimate that an industry association extrapolates from a survey of its member operators; the free summary gives no respondent count. It covers all telecoms fraud, of which fraud over VoIP is one part, and the summary prints only the top two methods and types.

How long do callers wait on tax helplines, and what do call-centre staff earn?

It depends on the helpline and on the measure. The UK and US tax agencies each publish their own answer rates and waits, defined differently, and US pay data covers customer service representatives in every setting, call centres included.

11:36 Apr 2026 11:26 May 2026 11:46 Jun 2026 09:38 Jul 2026 08:20 Aug 2026 08:30 Sep 2026
View data table
Average speed of answer on HMRC's telephone helplines by month, April to September 2026, in minutes and seconds. Monthly values can be revised between quarterly publications.
PeriodValue
Apr 202611:36
May 202611:26
Jun 202611:46
Jul 202609:38
Aug 202608:20
Sep 202608:30
Average speed of answer on HMRC's telephone helplines by month, April to September 2026, in minutes and seconds. Monthly values can be revised between quarterly publications. (HMRC (performance data, Q2 2026 to 2027), Nov 2026)
  • Callers to HM Revenue and Customs, the UK tax authority, waited on average 8 minutes 52 seconds for an adviser between July and September 2026, compared with over 13 minutes in the same period of 2025; 90.7% of callers who wanted an adviser were answered, against a target of 85%. (HMRC (performance update, Q2 2026 to 2027), Nov 2026)
  • From April to September 2026, HMRC's average speed of answer was 10 minutes 15 seconds, 89.3% of adviser attempts were handled, 42.7% of calls waited more than 10 minutes and the average call took 16 minutes 47 seconds to handle. (HMRC (performance data, Q2 2026 to 2027), Nov 2026)
  • HMRC received 13.542 million telephone contacts from April to September 2026: 9.218 million were attempts to reach an adviser, of which 8.230 million were handled, 4.321 million contacts were handled by automated systems and 0.991 million were not handled. (HMRC (performance data, Q2 2026 to 2027), Nov 2026)
  • Customer satisfaction with HMRC from April to September 2026 was 80.8% for digital services, 73.8% for webchat and 58.7% for phone, on its own customer survey. (HMRC (performance data, Q2 2026 to 2027), Nov 2026)
  • In the 2026 US tax filing season the Internal Revenue Service received 48.1 million calls across all its lines; telephone assistors answered 9.9 million (21%) and the average time on hold was 14 minutes. (IRS (National Taxpayer Advocate mid-year report release), Jun 2026)
  • In the prior filing season the IRS received 50.2 million calls, of which telephone assistors answered 12.4 million (25%), with an average wait of 8 minutes. (IRS (National Taxpayer Advocate mid-year report release), Jun 2026)
  • On its Accounts Management lines alone, the IRS reported a 73% Level of Service in the 2026 filing season, a measure that leaves out other lines, calls routed to automation and callers who hang up before entering a queue. (IRS (National Taxpayer Advocate mid-year report release), Jun 2026)
  • An audit body's interim figures to 28 February 2026 show assistors on the IRS Accounts Management lines had answered 3.2 million calls (4.5 million at the same point in 2025) with an average speed of answer of 8.2 minutes (3.4 minutes in 2025); the service-rate goal was cut to 70% for the 2026 filing season from 85% in 2025. (TIGTA (2026 filing season interim results), Jun 2026)
  • Customer service representatives held about 2.7 million jobs in the United States in 2025 (2,666,000), with median pay of $21.53 an hour, or $44,770 a year, in May 2025. (BLS (Occupational Outlook Handbook), Oct 2026)
  • US employment of customer service representatives is projected to decline 5% from 2025 to 2035, to 2,524,100 jobs, a loss of 141,800, while about 289,500 openings a year are still projected, on the labour statistics agency's projection. (BLS (Occupational Outlook Handbook), Oct 2026)
  • Median hourly pay of US customer service representatives in May 2025 was $22.47 at insurance carriers, $21.77 in professional, scientific and technical services, $21.49 in healthcare and social assistance, $17.96 in retail trade and $17.68 in business support services. (BLS (Occupational Outlook Handbook), Oct 2026)

Each figure keeps its own measure. HMRC's figures describe one UK government helpline operation, not UK call centres in general. For the IRS, the 21% is the share of calls on all lines answered by a person, the 73% Level of Service covers the Accounts Management lines only and excludes several kinds of call, and the audit body's figures run only to 28 February 2026; these are different measures and are not set against each other. The US pay and jobs figures cover customer service representatives in all settings.

Glossary

VoIP
Voice over internet protocol: phone calls carried as data over a broadband or mobile data connection.
Interconnected VoIP
The FCC's regulatory term for a VoIP service that can call and be called from the public telephone network. Calls that stay inside an app are not included.
Switched access line
The FCC's term for a traditional phone line connected to a circuit-switched exchange. Most run over copper, some over fibre.
PSTN
Public switched telephone network: the traditional circuit-switched phone network that VoIP is replacing.
Over-the-top (OTT) VoIP
VoIP sold by a company that does not supply the customer's broadband. Cloud phone systems for businesses fall here.
Form 477
The mandatory filing in which US voice and broadband providers report their lines and subscriptions to the FCC twice a year.
ILEC
Incumbent local exchange carrier: the historic local phone company in a US area.
Wireless-only household
In the US health interview survey, a home with at least one mobile phone and no landline. A VoIP home phone counts as a landline there.
Wire centre
A local telephone exchange building and the area it serves. AT&T states its copper retirement progress as shares of wire centres.
Voice over broadband (VLB)
The French regulator's term for a phone service delivered over a broadband access, as opposed to the switched network (RTC).
Tele-density
Telephone subscribers per 100 population, the measure the Indian regulator uses.
0AB-J and 050 numbers
Japan's two kinds of IP telephone number: 0AB-J numbers are ordinary geographic numbers, 050 numbers are non-geographic.
VoLTE
Voice over LTE: phone calls carried as data over a 4G mobile network.
Robocall
An automated call that plays a recorded or generated message. The private index cited here includes wanted calls such as notifications.
STIR/SHAKEN
The US caller-ID authentication framework, in which the originating provider signs a call so later providers can check the number was not spoofed.
Attestation (A, B, C)
The level of a STIR/SHAKEN signature. A means the provider knows the customer and their right to use the number; C means it only knows where the call entered its network.
Robocall Mitigation Database
The FCC's register of voice providers' certifications. Providers may not accept traffic from providers that are not listed.
Level of Service
The IRS's helpline measure for its Accounts Management lines, which excludes other lines, calls sent to automation and early hang-ups.

Methodology and sources

Figures come from the bodies that run each measure: the FCC's Voice Telephone Services reports and enforcement documents, the CDC's National Health Interview Survey, the regulators Arcep, Bundesnetzagentur, CNMC, TRAI, ACMA and Japan's communications ministry, the ITU, the FTC's Do Not Call data, HMRC, the IRS National Taxpayer Advocate, TIGTA and the Bureau of Labor Statistics, and the companies' own statements from AT&T, BT, RingCentral, Zoom, Microsoft, Bandwidth, Twilio, Five9, WhatsApp, YouMail and TransNexus, plus the CFCA's fraud loss survey. Each figure links to the document that published it, with the date, and source labels name the report. Measures are kept apart: VoIP subscriptions, traditional lines, mobile subscriptions and shares of homes without a landline are different counts, countries are not ranked against each other, company figures are not ranked, and switch-off dates are plans unless the closure is complete. For FCC line counts the newest edition is used and older editions are labelled. Where a producer printed two different values for one figure, the figure is left out. The research found no openable regulator line counts for the UK or Canada, no EU-wide totals, no figures for emergency calls from VoIP and no worldwide VoLTE counts, so the page has none. Forecasts and company guidance are labelled as such. The page is reviewed monthly.