Satellite internet in New Zealand: Starlink and what comes next

These are government test-panel figures, not provider claims. New Zealand is unusual in having them for satellite at all — and in still collecting them.

Commerce Commission, Measuring Broadband New Zealand Report 28 — published July 2026, testing conducted 1–30 April 2026. Figures are Table 3 exactly, not derived. Worth knowing the sample sizes: 68 Whiteboxes on Residential Max, 52 on Residential 100 Mbps, and only 27 on 4G fixed wireless in non-fibre areas — the Commission itself says to factor in the error bars on the smaller samples. Starlink is the only LEO provider in the programme.

What the Commission states directly, in its release of 29 June 2026: the 27% rural share, that Starlink is now the largest rural broadband provider, that most of the growth came from households leaving legacy copper, and the 54/46 split. The connection counts, growth rate, revenue figures and OECD ranking are from media reporting of the full Annual Telecommunications Monitoring Report (NZ Herald and others, June–July 2026) rather than read by us from the report itself — treat them as well-sourced but second-hand. The Commission publishes with a lag, so the share is likely higher again today.

Amazon Leo’s first-generation shells are optimised for latitudes 30–56°. New Zealand spans roughly 34°S to 47°S, so the whole country is inside it — from Cape Reinga to Bluff. That is a better position than Australia, where the tropical north sits outside the band and waits longer.

What is known about timing: the first New Zealand ground station is under construction and Amazon has advertised for local staff, but no launch date has been announced. Australia is expected to get service first. The Commerce Commission expects Amazon’s entry to slow Starlink’s momentum, and one rural ISP guide puts competition arriving somewhere between 2026 and 2028.

Latitude band from Amazon’s published first-generation shell design. Ground-station and staffing detail reported by the NZ Herald, June 2026. The Australian timeline is here, and it is further along.

This is where New Zealand is genuinely ahead of Australia. One NZ launched satellite texting first; Spark switched on its own Starlink-powered service on 1 April 2026 — so there are two carriers competing on satellite-to-mobile rather than one.

Prices are indicative add-on rates as reported mid-2026; confirm with the carrier. As in Australia, this is texting, not a full mobile service, and it is not a replacement for emergency calling — the same caveats apply.

More than a quarter of New Zealand’s rural broadband now runs through a single US-controlled constellation, and the country is the most satellite-dependent in the OECD per capita. That has become a live policy question rather than a hypothetical one.

In July 2026 the ISP association formally asked the Commerce Commission for a market study into Starlink, arguing a wholesale model could keep the benefits while retaining local accountability. WombatNET’s founder put the blunt version to RNZ in April: if Starlink were switched off tomorrow, it would be quite bad for New Zealand.

The Commission’s own view is more measured — it calls the prospect of being “switched off” remote, and says it becomes more remote as more satellite operators enter. It also notes something that gets lost in the headlines: when rural households leave copper, 54% choose Starlink and 46% still choose a local provider, valuing on-the-ground support over raw speed.

The Commission’s own commissioned adviser went further than the Commission has. In an independent report on telecommunications regulation, Richard Feasey warned there is “a material risk that Starlink would win the competition ‘for the market’ and become the dominant if not monopoly supplier of broadband services in rural areas”, and argued rural fibre deployment matters for social inclusion as well as competition.

And there is a concrete ask on the table that has nothing to do with satellites. WISP operators want the 6 GHz band opened for outdoor use, which would let land-based providers add capacity and give rural users an alternative. Radio Spectrum Management declined in 2022 and has kept a “watching brief”; the band is available for outdoor transmission in the United States and Canada. Operators say they will invest once it is approved. That is the clearest example of a domestic alternative being constrained by something other than Starlink.

Two other figures worth holding alongside the market-share number: Federated Farmers reports 36% of New Zealand farms now use satellite internet, and Starlink earned more than $100 million from New Zealand last year. Tech Users NZ chief executive Craig Young’s summary of the position: all our eggs in one basket.

Ispanz letter reported July 2026; Commerce Commission commentary June 2026; Feasey report commissioned by the Commission and reported by RNZ, May 2026; spectrum request via Farmers Weekly and WISPA, May 2026. We have no position on the market study — the figures are here so you can form your own.

The honest test is the same as anywhere: check fibre first. New Zealand’s UFB rollout is complete in urban areas, and where fibre exists it beats satellite on price, latency and consistency. Around 13% of New Zealanders cannot get it — that is who satellite is for.

If you are in that 13%, the Commission’s own numbers make the case better than any provider page: about 4.6× the rural 4G fixed-wireless average and 22× ADSL. Be careful with the “twenty times faster than copper” line you will see quoted — it holds against ADSL, but VDSL averages 63 Mbps, so the real gap there is closer to 4×. If you are on decent VDSL, satellite is an upgrade rather than a transformation. If there is a local WISP in your area, talk to them too — the 46% who stay with local providers are not making a mistake, they are buying support.

Hardware and plan pricing changes often and differs from Australia; check starlink.com for your address before ordering. Full New Zealand pricing and the tier question are here, and The Australian figures, for comparison, are on one page here.

Five companion pages, each on a question this one only summarises.