Paul Colmer, in an article for TechCentral, advocates that the Competition Commission reject ACT's application on behalf of its members for a five-year exemption concerning part of the rural network rollout. He meticulously argues his position, demanding a direct answer rather than merely a restatement of the existing stance.
Colmer acknowledges that co-location of equipment, site leasing, roaming, and open access infrastructure have their place. However, his objection stems from the fact that competitors are aligning their future plans. In his view, sharing already built equipment reduces losses, allowing each operator to independently decide where to compete.
The author notes that in the most peripheral parts of the country, realistic options are not numerous independent networks or one single shared network. Instead, what is needed is shared infrastructure capable of supporting competing services, or deployment that arrives too late or never becomes commercially viable.
The economic basis of the issue is not disputed: to serve the first customer, a site must be acquired, built, powered, secured, and connected to the backbone. In densely populated urban markets, several operators can cover these fixed costs and continue to compete. In sparsely populated areas, where every operator must overcome the same threshold to achieve a small revenue base, owning infrastructure can become a barrier to entry rather than a guarantee of competition. The requirement for every operator to replicate equally expensive assets before being allowed to compete is itself a market entry and expansion constraint.
Competition, Not Duplication
On this matter, the author disagrees with Colmer on fundamental principles, not details. The goal of competition policy is competition itself, not duplication. Operators compete on price, coverage, quality of service, network performance, innovation, products, and customer experience. Ownership of a tower or fiber optic route does not grant control over the retail market.
Infrastructure sharing differs from customer sharing, market allocation, or price coordination: the former concerns how connectivity is provided, while the latter concerns how operators fight for customers. Duplicated infrastructure is beneficial where it increases capacity, resilience, or real choice. This in itself is not proof of market competitiveness.
Colmer's warning is that in a thin market, the first operator to build may become the only one with a rational reason to be there, and exclusivity will arise without signing an exclusivity agreement. However, this argument is flawed because it ignores three factors:
- First—infrastructure sharing is already a requirement, not a benefit: access on open, fair, and non-discriminatory terms turns the first build into an entry point, not a closed position.
- Second—retail competition does not depend on tower ownership. Mobile virtual network operators compete for customers on networks they do not own, and one physical network can serve several competing providers.
- Third—Wapa members can use this infrastructure—small operators who have succeeded in marginal areas for years will gain access to assets they could not build themselves.
Agreements that restrict access, entrench the positions of incumbent players, or increase competitors' costs must still attract antitrust scrutiny. Neither benefit nor harm should be assumed solely because of sharing; both aspects must be proven.
South Africa's competition legislation already draws this line. Section 4(1)(a) of the Competition Act prohibits agreements between competitors that substantially lessen or prevent competition, unless a party can demonstrate technological, economic, or other pro-competitive benefits outweighing this effect. Section 10 provides a separate pathway to exempt certain agreements if they contribute to achieving legislative objectives—among which are effective market entry, participation, and expansion for small and medium enterprises and firms owned or controlled by historically disadvantaged persons. These two provisions perform different legal functions. Both are based on one premise: cooperation between competitors is not automatic suppression of competition, and the distinction is made on the basis of evidence.
Sharing is not neutral from a competition standpoint, and the author does not believe it is. Competitors remain competitors. Commercially sensitive information unrelated to the infrastructure's purpose must remain protected. Retail pricing, customer engagement strategy, product decisions, and market distribution must remain subjects of independent competition.
Risks are managed by establishing clear limits on scope, information barriers, aggregation, and anonymization where necessary, independent oversight, and compliance with competition law. Theoretical risk is not proven harm. Competition analysis requires evidence of likely consequences, market conditions, and the link between the action and the alleged harm. Considering possible risk as an inevitable outcome does not substitute for such analysis.
Regarding affordability, the author will be candid, as Colmer is entitled to a direct response. Improving deployment economics does not guarantee lower retail prices, and the author does not intend to claim that it does. Retail pricing must remain an area of active competition, and this is precisely what the current application excludes.
An Open, Evidence-Based Process
What ensures efficiency is something else: capital that is not spent repeatedly on the same duplicated assets can be directed towards expansion, increasing capacity, enhancing resilience, modernization, and coverage in areas where investment would otherwise be hard to justify. Coverage that people cannot afford does not ensure inclusivity. Just as an available service in a location where there is no network. Both problems must be solved, and neither is solved by insisting on duplication for duplication's sake.
This is how it can be phrased. It is not a matter to be resolved in journalistic articles, neither mine nor Colmer's. The Competition Commission's process exists to test exactly these assertions based on evidence and submitted materials. ACT supports an open, evidence-based process, and the author calls on Wapa, the smaller operators it represents, and any other interested party to present their arguments there, officially, where they can be weighed.
[ ]
