On September 9, 2026, Vietnam sold three 2x5 MHz pairs in the 900 MHz band. VNPT took C3-C3′ (900-905/945-950 MHz) for about VND 1.079 trillion. Viettel took C4-C4′ and Vietnamobile took C5-C5′, each for about VND 1.077 trillion. The starting price was roughly VND 1.057 trillion per pair, so combined bids passed VND 3 trillion. Licences run 15 years. According to the Ministry of Science and Technology, the bid increment was VND 10 billion per round.
This is a genuine first. Earlier Vietnamese auctions covered the 2.6 GHz, 3.7 GHz and 700 MHz bands. As VnEconomy reported, this is the first time Vietnam has auctioned a band that operators had actively used for years. The Authority of Radio Frequency Management described it as the first time a former 2G band had been re-planned for 4G and 5G and put to public auction.
The case for the state's approach
The strongest argument for Hanoi's design is that it lets a scarce public resource be repurposed in an orderly way. Sub-1 GHz spectrum is valuable because it carries far and penetrates buildings. That matters for rural, mountainous and island coverage, and for indoor signal in cities. An auction gives a transparent price, avoids the discretion of a beauty contest, and returns revenue to the budget. Sequencing also matters. The 2G shutdown was set for September 15, and the auction came six days before it. Operators could not have planned refarming without knowing who would hold which block.
The legal basis was put in place ahead of the sale. Circular 29/2026/TT-BKHCN, issued and effective May 31, 2026, designates 890-915 MHz and 935-960 MHz for IMT-2000, IMT-Advanced, IMT-2020 and later systems from September 16, 2026, meaning 3G, 4G and 5G, with 3G allowed only until September 14, 2028. The circular amends the 2024 band plan issued by the former Ministry of Information and Communications. Clear rules published months before bidding are the kind of predictability investors need.
What the numbers actually show
The outcome was not a price war. The three winning bids sit only slightly above the reserve. VNPT's winner, the highest, is roughly 2% above the starting price. The other two are only marginally higher. The bid increment was VND 10 billion, so each winner paid only a round or two above reserve.
Two design choices explain this. First, the circular caps any single organisation at 2x10 MHz across the band, so no operator could take more than two of the five 5 MHz pairs. Second, the auction offered three blocks, and three operators won one each. When supply roughly matches the number of serious bidders and each bidder is capped, there is little to fight over. The result is efficient in a narrow sense: each block went to a licensed operator that wanted it, at a price the state had judged fair. But it is not evidence of competitive price discovery, and the revenue figure should not be read as a measure of what the spectrum is worth.
MobiFone did not win a block. Vietnam News reports that, and the coverage I reviewed does not say whether it bid or why it did not win one. It would be speculation to explain the gap. The competition question is nonetheless real. A large operator with no sub-1 GHz block will face higher costs for indoor and rural coverage than its three rivals, unless it relies on other bands or on sharing.
Low reserve pricing is the right instinct
Spectrum fees are a cost that operators have to recover from consumers or from network investment. The higher the price, the less capital remains for base stations and backhaul. Evidence from around the world suggests that revenue-maximising auctions can leave coverage worse. Vietnam's near-reserve outcome, whatever its cause, means about VND 3 trillion went to the budget and the rest of the operators' capital stays available for deployment. The obligation attached to the licences, as the ministry describes it, is to deploy 4G and subsequent technologies. That is a reasonable, technology-neutral condition, and it avoids prescribing coverage targets that could distort investment.
There are risks. The circular allows 3G in the band only until the end of September 14, 2028. Two years of overlap between legacy 3G and new 4G/5G carriers in a narrow 5 MHz channel is tight. The 2G exit is also not painless. In 2024, operators stopped serving GSM-only handsets, but Telecom Review Asia notes that early 3G and 4G phones lack VoLTE, which is why voice on 2G was kept until September 2026. Users of low-cost feature phones in remote areas are the most exposed, and the state should track whether they are stranded rather than assume the transition is complete.
What regulators should do next
Three lessons follow. First, publish the reasoning for the per-operator cap. Caps are defensible for keeping the market plural, but the ministry should say whether it expected the near-reserve result, because bidders and future auction designers need to know. Second, when the next band comes up, set supply and caps so that they do not settle the outcome before bidding begins. The 470-694 MHz band, where Circular 27/2026/TT-BKHCN has set a conditional path from television to mobile use, will be a harder test, since it involves incumbent broadcasters. Third, keep fees modest and tie them to deployment. The goal of spectrum policy is coverage and capacity that reach people, not the auction total.
Vietnam has done the difficult part: it retired a legacy network on a published timetable, changed the band plan in advance, and sold the freed spectrum within days. The auction result should be judged over the 15-year licence term, not by the size of the cheque.