Summary

  • China Tower Corporation Limited's three operator-shareholders signed framework agreements on 18 January 2023 that run to 31 December 2027, and the price schedule inside them makes an additional tenant cheaper than the schedule it replaced.
  • At a two-tenant site the second, non-anchor tenant pays 67.6% of base price, down from 70% under the schedule effective 1 January 2018; at a three-tenant site the rate is 57.6%, down from 60%.
  • A rising tenancy ratio therefore records volume won, not pricing power earned. The price of the marginal tenant is set by the customers who control the seller.

China Tower Corporation Limited owns steel and power connections, and its largest customers hang radio equipment on them. The company's directory record is published here (https://btw.media/en/directory/china-tower-corporation-limited-cn). The industry scores that business with a single number — tenants divided by sites — and on that measure the company looks like a machine for extracting more from assets already built. A secondary aggregator's account of the first half of 2026 gives 2.172 million sites, 3.871 million tenants and a tenancy ratio of 1.78, against revenue of RMB 48.69 billion, of which RMB 35.26 billion came from the tower business (https://telecomlead.com/telecom-services/telecom-tower-economics-2026-china-tower-indus-american-tower-sba-and-helios-compared-on-tenancy-revenue-and-profitability-128277). Those are one aggregator's figures; no issuer-filed interim announcement for the period is in hand for this account, so they carry secondary weight.

The ratio counts tenants. It does not price them. That price is written in framework agreements with three companies that are at once the biggest customers and, through their shareholdings, the controllers of the seller. Twice in the past decade the renewal has moved the price of an extra tenant in the same direction.

Two schedules, one direction

The agreements in force were signed on 18 January 2023, run from 1 January 2023 to 31 December 2027, and are described in China Tower's 2025 annual report, which sends readers to a 15 December 2022 circular for the tariffs (https://doc.irasia.com/listco/hk/chinatower/annual/2025/ar2025.pdf). The circular carries the table that matters. A non-anchor tenant takes a 32.4% discount to base price on a two-tenant site and 42.4% on a three-tenant site, leaving 67.6% and 57.6% payable; both non-anchor tenants on a three-tenant site get the three-tenant rate. Anchor tenants pay 62.6% and 52.6% of base price. Site and power-access charges are discounted on a separate scale, at 60% and 50% payable for non-anchors and 55% and 45% for anchors (https://www1.hkexnews.hk/listedco/listconews/sehk/2022/1215/2022121500109.pdf).

China Telecom Corporation Limited, one of those three shareholders, wrote the same tariff as a formula in its own announcement dated 13 December 2022: 0.676B + 0.60(S+E) where a site holds two tenants, and 0.576B + 0.50(S+E) where it holds three, with B the base price and S and E the site and electricity-connection costs (https://www1.hkexnews.hk/listedco/listconews/sehk/2022/1213/2022121301435.pdf).

The schedule it replaced is visible in the July 2018 Global Offering prospectus: a 30% base-price discount for the second tenant on a two-tenant site and 40% for both the second and third tenants on a three-tenant site, effective 1 January 2018, giving payable fractions of 70% and 60% with anchor discounts of 35% and 45% (https://www.hkexnews.hk/listedco/listconews/sehk/2018/0725/LTN20180725011.pdf). China Mobile Limited filed the same step from the buyer's side. Its supplementary agreement, dated 31 January 2018, raised the two-lessee base-price sharing discount from 20% to 30%, set other lessees at discounts of 30% and 40% against 35% and 45% for the first occupier, took effect on 1 January 2018 and stated a term ending 31 December 2022 (https://www.sec.gov/Archives/edgar/data/1117795/000119312518133541/d510208dex426.htm).

Set end to end, the marginal tenant on a two-tenant site paid 80% of base price under the discount level that filing replaced, 70% from 1 January 2018 and 67.6% from 1 January 2023. Every payable fraction in the two published tables — anchor and non-anchor, two tenants and three — shifted by exactly 2.4 percentage points at the 2023 renewal. That is arithmetic on disclosed schedules, not a company-reported metric, and the honest reading is unflattering to a dramatic story: the 2018 cut was ten points, the 2023 cut barely moved a tariff. The direction, however, did not change.

Why the ratio cannot carry the argument

Co-location is a forgiving business. The structure is standing, the power connection is sunk, and a second or third tenant adds revenue against a modest incremental cost. That is my inference from sunk-cost logic, and it should be labelled as inference: the retained record supports a statement about the price of an additional tenant, not about the margin on one.

The monthly mechanics explain why the seller cannot bank a rising ratio. China Unicom (Hong Kong) Limited told the US Securities and Exchange Commission in June 2017 that the co-tenancy discount rate applies to each monthly payment according to the sharing situation at that time, with larger discounts for three operators than for two, starting when the incoming tenant's service begins and limited to facilities actually shared (https://www.sec.gov/Archives/edgar/data/1113866/000119312517199567/filename1.htm). Price resets with occupancy instead of being fixed at build. The count itself is a contractual object too: under the frameworks, operators jointly sharing networks and antennas may count as one.

So a tenancy ratio rising from one period to the next tells a reader that co-location happened. It cannot tell a reader whether the company kept the value, because the price of the increment is negotiated with the counterparties who own the company. On a site where two tenants now share, the second one's contribution was fixed by a schedule its own controlling shareholder helped set.

Where the record stops

Four blanks bound the conclusion. Revenue per tenant is not published separately from revenue per site. The incremental cost of an additional tenant is not disclosed at all. The size of a grandfathered class is unknown: parties already sharing acquired towers before 31 October 2015 pay 27.6% of base price and 30% of site cost, and the same China Telecom announcement states that the product price does not change when a third tenant joins — a cohort that cannot be monetised through densification, at a weight the documents do not give. And pricing for tenancies from anyone other than the three operators is absent from the record.

What the arithmetic can say is how little each site yields at the top line. RMB 35.26 billion of tower-business revenue spread across 2.172 million sites is roughly RMB 16,200 per site for the half, about RMB 2,700 a month. That is my calculation from two aggregator figures, not a company metric, and it deliberately ignores the three non-tower lines in the same report.

The 2027 test

The current framework expires on 31 December 2027. That date is the falsifiable test of everything above. If the next set of agreements holds the payable fractions at 67.6% and 57.6%, or lifts them, the pattern of renewal-by-discount is broken and this reading is wrong. If a third renewal trims them again, the tenancy ratio will keep rising while the price of each increment keeps falling, and the market will keep congratulating the company for the first number while ignoring the second.

China Tower is not a weak business. It is a business whose best customers write its price list, and the ratio the industry celebrates was never designed to measure that.