On 19 September, Circular 39/2026/TT-NHNN took effect. The State Bank of Vietnam now allows non-resident foreign credit institutions to use foreign currency and Vietnamese dong accounts at licensed Vietnamese banks for international payments and money transfers, on behalf of their own customers, under a written bilateral agreement with the host bank.
Headlines compressed that into one sentence: Vietnam opens the door to foreign payment companies. That reading is wrong, and it is the kind of wrong that costs companies a year. A foreign bank, a payment infrastructure provider, and a non-bank fintech stand in three completely different positions relative to this circular. None of them share a door. The useful question is not who is allowed in. It is whose money can actually cross the border, land, and reach a Vietnamese counterparty without breaking a rule somewhere in the chain.
What the circular actually does
Circular 39 amends Circular 16/2014, which has governed accounts of residents and non-residents at Vietnamese banks for over a decade. The old text covered account opening for non-residents reasonably well. What it never spelled out was whether a foreign credit institution could route third-party customer flows through those accounts as a settlement node. Banks operated in that grey zone by convention, with compliance teams on both sides unsure whether the arrangement was inside or outside the framework.
The new Article 2a closes the gap. Two conditions attach. First, a written agreement must exist between the foreign institution and the Vietnamese licensed bank before any customer payments flow. Second, every transaction must comply with Vietnamese foreign exchange law, non-cash payment rules, and AML obligations. The amended Article 9 puts the legal responsibility for those flows on the Vietnamese bank, and it cannot contract that liability away to the foreign counterparty.
The quiet half of the story is the dong. Foreign currency correspondent accounts have existed in practice for years. A VND correspondent account, letting a foreign bank receive, hold, and remit dong on behalf of customers, removes the USD intermediation step that VND trade flows have needed until now. That is structural, not cosmetic. It sits alongside the broader stack Vietnam has been assembling: the international financial centre framework, outbound FX rules, the digital asset pilot under the Law on Digital Technology Industry. Vietnam is building pipes, deliberately, piece by piece.
Three positions, not one queue
Now the part most commentary skips. Read Article 2a as a status provision, not a functional one. The right attaches to being a foreign credit institution, a bank or licensed deposit-taking equivalent. It does not attach to what your product does.
The foreign bank is the direct beneficiary. A Korean, Singaporean, or UAE lender with existing correspondent relationships can now negotiate an explicit agreement, formalise what was previously convention, and route VND and FX customer payments through a Vietnamese partner bank. The work is legal documentation and compliance architecture, not a licensing battle. If a bank is evaluating Vietnam as a settlement node for regional corridors, the account structure is now black-letter law.
The payment infrastructure provider is in a different spot. A stablecoin rail, a cross-border network, an e-money institution licensed in Singapore or Lithuania, none of these are credit institutions under Vietnamese reading. Article 2a gives them nothing directly. Their money can still move, but as a customer of a bank, not as a rightsholder. The correspondent agreement sits between two banks, and the rail sits outside it, contracting for services. That can work. It means the compliance burden lands on the bank that holds the funds, and the bank now carries legal liability it cannot pass back. Expect Vietnamese banks to price that risk into onboarding, and to be picky.
The non-bank fintech is nowhere in this circular. If you move or hold customer funds in Vietnam, you are in the intermediary payment services regime under Decree 52/2024: a local licensed entity, charter capital, fit-and-proper checks on your officers, level 3 information security certification for your systems. Circular 39 changes none of that. A fintech that read the headlines as a shortcut into Vietnam has misread the instrument by two licensing regimes.
Trace the money, then argue
The fastest way to see the three positions is to draw one payment end to end. Take a concrete case: a Singapore company needs to pay a Vietnamese supplier 500 million dong.
The sender debits Singapore dollars from its account. Someone must convert. If the sender's bank is a credit institution with a VND correspondent account under Circular 39, conversion and settlement can happen inside one agreement with one Vietnamese partner bank. That is the cleanest path the new rule creates.
If the sender is a fintech, the funds still move, but the fintech cannot be the settlement party in Vietnam. The money must land at a licensed Vietnamese institution that holds it, converts it under FX rules, and pays out. That institution signs the contracts, bears the Article 9-style liability, and owns the compliance file. The fintech owns the customer experience and the API. The licence follows the funds, not the software.
At the receiving end, the supplier gets paid into a Vietnamese bank account or an e-wallet attached to one. The wallet, if a non-bank provider runs it, exists because that provider holds an SBV licence. Every leg of this flow has a party that holds money, a party that signs, and a party that carries a licence or a bank. When you lay it out this way, the circular's actual scope becomes obvious: it shortens one leg for one category of institution. Everything else was already determined by rules that predate it.
There is a real-world parallel here. India built UPI and then spent years negotiating which foreign entities could plug into it, and the answer kept depending on status and reciprocity agreements, not on the technology. Nigeria's model routes everything through licensed local partners. The pattern repeats because it is the pattern regulators trust: local liability, local visibility, local money. Vietnam is not inventing a new philosophy. It is formalising the same one, account by account.
What this means for anyone planning entry
The practical conclusion is narrow and worth keeping. Circular 39 is good news if you are a bank, useful context if you are a rail, and irrelevant to your licence if you are a fintech touching funds. In all three cases, the binding constraint on operating in Vietnam was never the right to send money across the border. It is the landing structure inside the country: which entity holds funds, under whose licence, with which bank agreement, and who answers to the SBV when a transaction goes wrong.
That is why the first work of market entry here is not a pitch deck. It is a diagram. Draw the payment, leg by leg, and name the holder, the signatory, and the licensee at each step. The gaps show up immediately, and so does the partner you actually need. Most companies discover this after they have committed to a structure that looked fine in a slide and fails on the second transaction.
The circular opened a route, and for one class of institution it is a real route. For everyone else, it is a reminder that in this market, the money decides who you are.
Share this post