Malaysia Repriced 5G Twice. The Bill Lands on Two Balance Sheets

HONG KONG, Oct 31, 2026 – (ACN Newswire) – On paper, Malaysia finished its 5G argument. Two wholesale networks now run side by side, a state-owned wholesaler sells capacity to three tenants, and a fourth operator built its own. Subscriptions passed 30 million, adoption passed 88%, and the regulator spent 2026 rewriting the access rules to keep the two networks from strangling each other. Every milestone the policy promised has been met.

In practice, the people who paid for those milestones have not finished paying. CelcomDigi and Maxis, the two listed operators still buying wholesale capacity from Digital Nasional Berhad, are about to consolidate that wholesaler onto their own books as a 33% associate in the fourth quarter of 2026. They are doing so after watching its two largest customers leave. Telekom Malaysia served notice on 24 February 2026 and signed a three-year wholesale deal with U Mobile instead. U Mobile migrated every customer off DNB on 1 July 2026 and now runs its own network.

The judgment this article defends is narrow enough to be falsified: Malaysia’s mobile cost per gigabyte is falling for reasons that have nothing to do with retail competition, and the savings will be competed away before shareholders see them, because the two variables doing the work are both cost-side and neither one creates pricing power. If wholesale revenue per gigabyte stops falling in 2027, that judgment is wrong.

Two Cost Curves Moved at Once, and Only One Was Scheduled

The first variable is integration. CelcomDigi’s merger, closed in 2022, has been converting two networks into one on a timetable that keeps slipping at the edges. Network integration passed 90% by May 2026. The IT layer, the billing and customer relationship systems that decide whether a merged company behaves like one company, was 85% done in February 2026 and is now scheduled to finish in the second half of 2027.

The money is real and already showing. CelcomDigi booked RM40 million of savings in the first quarter of 2026 and RM141 million across the first half, then raised its full-year cost target to RM470 million from RM450 million. Management still guides to RM700 million to RM800 million of annualised savings once integration is complete after 2027. Those are operating expenses and cost of goods sold, not one-off accounting gains.

The second variable is wholesale. DNB added 100MHz in the 3.3GHz to 3.4GHz band at the start of 2026, which combined with its existing 3.4GHz to 3.5GHz holding gives it a contiguous 200MHz block, the largest mid-band footprint in the country. Doubling capacity at a fixed site count halves the cost per gigabyte carried, if volume arrives. U Mobile’s exit and TM’s switch mean the volume that has to carry that fixed cost is now spread across three tenants instead of five.

These two variables are independent. One is an internal project with milestone dates. The other is a contract and spectrum story decided by the regulator and by a rival’s build. They do not share a cause, and that is why they stack rather than offset.

A Rulebook Rewritten to Keep Both Networks Alive

On 1 August 2026 the Commission Determination on Access List, Determination No. 3 of 2026, replaced the 2021 determination and kept two wholesale 5G services on the regulated list: 5G Standalone Access, and 4G Evolved Packet Core with 5G Radio Access Network Access. Both are radio-level products. Neither includes access to the provider’s core network, which means an access seeker brings its own core and rents only the antennas.

The detail that matters commercially is what MCMC did not do. It declined to write an explicit permission or an explicit ban on DNB and U Mobile using the regulated framework to obtain multi-operator core network access to each other’s radio. It also declined to exclude either wholesaler from becoming an access seeker of the other. In plain terms, the two rivals can rent each other’s towers where building a second radio site costs more than the regulated rate, and the regulator said so on the record: deploying another 5G RAN in a given area can cost substantially more than obtaining regulated access.

MCMC described the resulting wholesale market as a duopoly, and used that word to justify heavier oversight of network planning and change management. A duopoly with a regulated price floor and a mutual access option is not a competitive market in the ordinary sense. It is a managed one. For an operator buying capacity, that is good news on price and bad news on flexibility, because the rate is set by a determination rather than negotiated quarter to quarter.

The Challenger Rents Its Backbone From a Rival

There is a circularity in the Malaysian arrangement that rarely makes it into the coverage. U Mobile built its own 5G radio network and reached more than 85% population coverage by mid-2026, which is the fastest rollout the country has recorded. It did not build the fibre underneath it.

In May 2025, U Mobile signed a ten-year agreement with Telekom Malaysia valued at RM2.4 billion for leased fibre to its radio sites, data centre and edge facility capacity, and inter-regional trunk connectivity. TM operates roughly 740,000 kilometres of fibre, the largest footprint in the country. So the challenger’s radio network runs on the incumbent fixed-line operator’s ducts, and that same fixed-line operator became the challenger’s first major wholesale customer in February 2026.

The arrangement is rational for both sides. TM monetises ducts it already owns and escapes a wholesale contract it no longer wanted. U Mobile gets national backhaul without a decade of trenching. The cost is that neither has a clean hedge: if U Mobile’s retail pricing collapses, TM’s ten-year contract is only as good as its customer’s balance sheet, and if TM raises backhaul rates at renewal, U Mobile’s unit cost advantage narrows.

That is the second reason the cost story is not a simple deflation narrative. Radio costs per gigabyte are falling because spectrum doubled and sites are shared. Backhaul costs are contracted for a decade at a fixed structure. The blended unit cost moves more slowly than the radio number suggests, and the part that moves slowest is the part neither operator controls.

What the Second Quarter Reported

The February 2026 price rises were the tell. CelcomDigi, Maxis and U Mobile lifted postpaid and prepaid prices by 5% to 10% in the same window, which is not what a price war looks like. Sector service revenue rose 2% year on year in the second quarter, and sector core net profit still fell 6.6% to RM1.44 billion, down 19.4% from the first quarter, according to UOB Kay Hian’s tally of the four listed operators.

Read that pairing carefully. Prices went up, revenue barely moved, and profit fell. The cost of holding a subscriber is rising faster than the price the subscriber pays, even after a coordinated increase.

Table 1: How the two listed mobile operators performed on the metrics that decide unit economics, second quarter 2026

Source: MBSB Research via Business Today 🔗, The Edge Malaysia 🔗

CelcomDigi’s postpaid base grew by 191,000 subscribers to 6.1 million, and its postpaid ARPU still fell 4.5% to RM60. That combination, more subscribers at a lower average, is what a market looks like when the marginal customer is being bought with a cheaper plan rather than won with a better network. Maxis held postpaid ARPU at RM70.50, down 1.1%, and lost 210,000 prepaid users.

The prepaid side is where the arithmetic bites hardest. CelcomDigi’s prepaid revenue fell 3.2% to RM1.016 billion, and its prepaid ARPU held at RM28 only because the cheapest customers left first. Malaysia’s mobile penetration was already 142.9% in the first quarter of 2026, and all four of the regulator’s penetration measures fell quarter on quarter, with prepaid down 2.1% and SMS down 5.7%. There is no unserved population left to add. Every prepaid subscriber an operator gains is one another operator lost.

One Line Item Does Not Fit the Story

Home fibre is the exception, and it explains why the two survivors look so different. CelcomDigi’s fibre revenue rose 47.2% to RM95 million with ARPU up 8.4% to RM103, off a small base of 315,000 subscribers. Maxis booked RM252 million of fibre revenue, down 0.8%, at an ARPU of RM109.10. Telekom Malaysia’s unifi did better than either, with revenue up 5.3% to RM1.45 billion and ARPU up 4.8% to RM132.

Fixed line is the only line item in this sector where price is not collapsing, because the product is sold per household rather than per gigabyte. That is also why the mobile-only story is the fragile one: an operator selling only mobile has no second product to bundle, and bundling is the only remaining defence against churn.

Axiata illustrates the exposure. Its core net profit fell 74.1% to RM42.5 million in the second quarter of 2026, dragged by ringgit translation and withholding tax, and it carries an ownership stake in CelcomDigi rather than a retail mobile business of its own. At its February 2026 analyst day, 53% of the roughly 50 attendees polled named DNB’s 5G wholesale pricing as the biggest risk to the investment case, while 77% named asset monetisation as the thing that would move the share price. Both answers point at the same balance sheet, from opposite directions.

Three Shareholders Inherit a Wholesaler That Lost Its Tenants

The consolidation scheduled for the fourth quarter of 2026 puts DNB onto CelcomDigi’s and Maxis’s books as a 33% associate each, with YTL Communications holding the third share. The Ministry of Finance exercised its put option to exit, and the three operators bought the government’s stake at roughly RM327.9 million each.

The asset they are buying has a thin recent history. DNB lost RM1.2 billion in the financial year ended December 2024 and carried about RM4.9 billion of liabilities. It refinanced a RM1.5 billion Ministry of Finance loan, which is what made the consolidation possible without a cash call. Analysts project a net loss of about RM800 million for 2026, narrowing to roughly RM400 million in 2027 as the revenue base doubles.

Here is the tension. The 2027 recovery assumes revenue doubles while capacity has also doubled. Those two facts push in opposite directions on price. If DNB fills the new 200MHz with traffic from three tenants, its cost per gigabyte falls and it can cut wholesale rates without cutting revenue. If it cannot fill the spectrum, it has to cut rates anyway to attract volume, and the revenue doubling does not happen. BIMB has estimated the drag at about RM400 million a year, equal to 20% to 25% of downside for CelcomDigi and Maxis.

None of this is visible in a retail price. It sits in a wholesale contract that a consumer never reads.

Who Gets Hit Before the Next Set of Results

Three groups feel this first, and none of them are the two listed incumbents.

· Mobile virtual network operators renting capacity without a fixed-line bundle. Their input cost falls with wholesale rates, but so does the retail price they can charge, and they have no fibre to fall back on.

· Tower and site landlords whose contracts were priced against a four-tenant market. Two tenants consolidated and one built its own network.

· Any operator that chose to build rather than buy. U Mobile raised RM4.3 billion in syndicated financing and reached more than 85% population coverage, which is a genuine achievement and also a fixed-cost commitment that must be serviced at whatever retail price the market settles on.

The incumbents are protected by a different mechanism: they own the wholesaler that supplies them, so a wholesale price cut is partly a transfer from one of their own subsidiaries to another. That is a comfort for the income statement and a problem for the return on capital, because the money is being moved rather than earned.

Enterprise Is the Only Line Where Price Holds

Consumer revenue is a volume business with a shrinking volume. Enterprise revenue is a contract business, and the second quarter of 2026 showed the difference in the same reporting period.

CelcomDigi’s enterprise solutions revenue rose 19.7% year on year, credited to cybersecurity and digital contract wins, while its enterprise mobile revenue fell 9.9% on lower bulk SMS traffic. Maxis grew enterprise revenue 3.7%. TM One added 3.3%, helped by public sector demand. The pattern is consistent: the connectivity component of enterprise is deflating, and the services component sitting on top of it is not.

That matters for the cost argument because services revenue carries a different cost structure. A gigabyte costs the same to carry whether it belongs to a consumer or a hospital group, but the hospital group pays for integration, security and a service level agreement on top. Those add-ons are priced against labour and software, not against spectrum. An operator that cannot hold consumer ARPU can still hold enterprise margin, provided it sells something other than the pipe.

The catch is scale. CelcomDigi services 22,000 retail touchpoints built for a consumer business that is shrinking, and it is spending the integration budget on billing and CRM systems rather than on the sales motion enterprise requires. The RM470 million savings target is a cost programme, not a growth programme, and the two compete for the same management attention in the same financial year.

The Prepaid Base Is Not Coming Back

Malaysia’s prepaid decline is often read as a migration story, customers graduating to postpaid. The numbers do not support a clean reading. Postpaid ARPU is falling at both operators while postpaid subscriber counts rise. If migration were the driver, the incoming cohort would be lighter users with lower willingness to pay, which is exactly what an entry-level postpaid plan looks like.

What is happening instead is that the prepaid product lost its reason to exist. Prepaid used to be the only way to get a data connection without a contract and a credit check. It is now one of several, and the alternatives include a fibre bundle at home, a second SIM inside a family plan, and a cheap postpaid tier that costs less per gigabyte than the prepaid top-up it replaced.

MCMC’s first-quarter figures show the ceiling clearly. At 142.9% penetration, the country has more SIM cards than people, and every penetration category fell in the same quarter. A market that has run out of new users can only grow revenue by raising what each user pays. The February 2026 increase tested that, and the second quarter answered: revenue up 2%, profit down 6.6%.

Signals That Would Break This Argument

The argument here is that cost deflation is outrunning any ability to keep it, so the benefits go to customers and the balance sheets absorb the difference. Three observations would prove it wrong.

First, if DNB reports wholesale revenue per gigabyte that is flat or rising in 2027 while carrying double the capacity, then the spectrum expansion did not translate into unit cost relief and the operators’ cost curve is not bending the way this piece assumes.

Second, if CelcomDigi’s postpaid ARPU recovers above RM65 without a corresponding drop in subscribers, price power exists and the February 2026 increase was the beginning of a repair rather than a one-off.

Third, if the integration savings arrive on schedule and are retained rather than passed through, sector margins should expand in 2027 even with flat revenue. Watch the gap between the RM700 million to RM800 million annualised guidance and reported operating expenses. If that gap closes and margins do not move, the savings went to customers.

There is a version of this market in which two wholesale networks, a merged incumbent and a self-built challenger produce genuinely better service at a lower price. That version is good for Malaysia and indifferent for the people who own the shares. The 30 million subscriptions are already there. What is not settled is who ends up paying for the network that carries them.

Data basis

Data as of October 2026. Figures are drawn from company filings, broker research and regulator publications covering the period from January 2025 to September 2026. Operator metrics for the second quarter of 2026 come from CelcomDigi’s Bursa Malaysia filing of 14 August 2026 and MBSB Research’s sector note of 8 September 2026, which reports postpaid and prepaid ARPU on a monthly basis and defines the prepaid subscriber change as the net reduction in active users during the quarter. Sector profit and the DNB projections come from UOB Kay Hian’s 2Q26 telecommunications review, which aggregates four listed operators and treats DNB as a 33% associate from the fourth quarter of 2026. Wholesale access terms come from the Commission Determination on Access List, Determination No. 3 of 2026, in force from 1 August 2026. Capacity figures are DNB statements as reported in January 2026. Savings guidance is company guidance, not an independent estimate, and the RM700 million to RM800 million range is annualised and post-2027.

Limitations

Three items would change the arithmetic. The DNB loss projections are analyst forecasts, not disclosed results, and the 2027 narrowing depends on revenue the company has not yet booked. Subscriber counts mix active and registered users across operators, so prepaid losses are not strictly comparable between CelcomDigi and Maxis. And the integration timeline has already moved once, from 2026 to the second half of 2027, which means the savings curve may be back-loaded beyond the window examined here.

Sources

· Malaysian Wireless 🔗, MCMC keeps 5G access open between DNB and U Mobile, August 2026

· Business Today Malaysia 🔗, telcos could see full financial impact of DNB in 2027, 8 September 2026

· The Edge Malaysia 🔗, CelcomDigi 2Q profit slides 9.3%, 14 August 2026

· UOB Kay Hian 🔗, Telecommunications 2Q26 review, September 2026

· The Exchange Asia 🔗, TM exits DNB, teams up with U Mobile for 5G, February 2026

This article was prepared by the expert team at Global Infinity Labs, drawing on insights and data provided through its MCP data services.

About Global Infinity Labs

Global Infinity  Labs is an AI innovation company specializing in MCP data services.

Email: support@globalinfinitylabs.com 

Copyright 2026 ACN Newswire. All rights reserved. www.acnnewswire.com