VAVaultgridplus
Guides › Compliance

Building a Digital Bank in Malaysia: BNM Licensing, RMiT, Platform Costs and Timeline

A digital bank in Malaysia starts under a foundational-phase asset cap and must meet BNM's technology risk rules before launch. Here's what that means for platform choice, budget and a timeline of roughly 30 to 40 months.

Vaultgridplus editorial team · November 2026 · 10 min read

What the BNM digital bank framework lets you build, and what it caps

If you're planning a digital bank in Malaysia, the first constraint isn't technology. It's the licence. Bank Negara Malaysia's (BNM) Licensing Framework for Digital Banks, issued at the end of 2020, set the rules for the first batch of licences. In 2022 BNM approved five applicants, under either the Financial Services Act 2013 (conventional) or the Islamic Financial Services Act 2013 (Islamic). The groups behind them included consortia built around Grab and Singtel, Sea and YTL, Axiata and RHB, AEON, and KAF. Most went live between late 2023 and 2025.

Three features of the framework drive almost every downstream decision:

  • A foundational phase. For its first years, typically three to five, a new digital bank operates under a total asset cap on the order of RM3 billion, a small fraction of what a mid-sized incumbent carries. BNM reviews progress before lifting the cap. Your platform has to be fully bank-grade from day one, even though your balance sheet stays small for years.
  • Tiered capital. Minimum capital funds are set at a lower level during the foundational phase and a higher level afterward. The figures commonly cited are RM100 million rising to RM300 million; RM100 million is roughly USD 21 to 24 million at recent exchange rates, which move. That capital absorbs losses and funds operations. It isn't a technology budget, so a vendor stack that consumes most of it is a warning sign.
  • A value proposition BNM can assess. Applicants had to show how they would serve underserved or unserved segments, such as gig workers, micro-SMEs and thin-file borrowers, and do it sustainably. That shapes product scope, and product scope shapes the platform.

BNM also expects Malaysian control. A Malaysian-led controlling shareholder or consortium is the norm, though foreign partners can hold significant minority stakes. Shareholders are assessed for fit-and-proper standing and ability to inject capital.

Advertisement

Licence windows don't stay open. BNM ran one application window and indicated it would consider further licensing after evaluating the first cohort. If you aren't licensed, your realistic routes are waiting for a future window, partnering with a licensee, or taking a stake in one. Confirm the current position with BNM and Malaysian regulatory counsel before committing budget to an application.

How RMiT shapes your cloud and platform architecture

BNM's Risk Management in Technology (RMiT) policy document is the technical rulebook every Malaysian bank operates under, digital or not. It has been revised several times, including changes that affected public cloud adoption. For a new entrant, RMiT is the main reason a cheap, fast, off-the-shelf stack often doesn't work.

Requirements that change platform choices

  • Availability for critical systems. RMiT expects very limited unplanned downtime for critical systems and tight recovery time objectives, measured in hours. Your core, payments switch and mobile channel need active-active or warm-standby designs across separate zones or data centres. A vendor SaaS service level of 99.5% won't meet that bar.
  • Cloud adoption. A bank putting critical systems on public cloud must assess the risk against BNM's cloud expectations, obtain board oversight and engage BNM as the policy requires. Expect detailed evidence on exit strategy, key management, concentration risk and audit rights. Check the current text for the exact process.
  • Independent assessment before go-live. New critical systems and major changes need an independent review of security and resilience before production. Plan it as a separate stage of several weeks.
  • Cyber operations. Expect a 24/7 security operations centre, DDoS mitigation, regular penetration testing, and prompt incident reporting to BNM. A senior officer accountable for information security, usually a CISO, is expected.
  • Customer authentication. Strong authentication for high-risk transactions, plus device binding and controls against mule accounts and scams. Expectations have tightened as BNM has pushed the industry away from SMS one-time passwords.

Cloud region and data residency

AWS, Microsoft Azure and Google Cloud have all launched or announced Malaysian regions, so the old trade-off between in-country hosting and hyperscaler tooling has largely gone. A common pattern is primary hosting in a Malaysian region, with disaster recovery in a second zone or a second provider. Workloads that BNM or your board treat as most sensitive stay in-country. Malaysia's Personal Data Protection Act was amended in 2024, with changes covering breach notification, data protection officers and cross-border transfers. Your counsel should review where analytics, fraud models and vendor support teams touch customer data.

BNM's outsourcing requirements apply alongside RMiT. Material arrangements, which can include core banking SaaS and a managed SOC, need due diligence, contractual audit rights, and in some cases BNM approval or notification. Find out early which of your vendors fall into that category.

Choosing core banking and the digital front end in Malaysia

The core banking decision comes down to three questions. Can the vendor show an RMiT evidence pack? Can it run in a Malaysian cloud region with the availability you need? Can it support Islamic products, if you want a dual or fully Islamic proposition?

The realistic vendor categories

  • Cloud-native cores such as Thought Machine, Mambu and 10x Banking. They configure quickly, are API-first and suit a lean product set. Check local implementation partners, Islamic product support (often built through configuration rather than delivered out of the box), and whether the SaaS or self-hosted model fits your cloud approach.
  • Established global cores such as Temenos, Oracle FLEXCUBE, Infosys Finacle and TCS BaNCS. Their Islamic banking modules are mature and they have local delivery teams. The trade-off is heavier implementations and higher customisation cost.
  • Regional players with a strong Southeast Asian presence, such as Silverlake. They know the local market and payment rails deeply. Assess how cloud-native their architecture really is.

For most foundational-phase digital banks, a cloud-native core paired with a strong local systems integrator is the better fit. The asset cap keeps scope narrow, typically savings, a debit card, personal financing and micro-SME lending, so speed of change matters more than feature breadth. The exception is a bank with a core Islamic proposition. There, a vendor with proven Shariah-compliant product engines, including profit calculation, tawarruq or murabahah flows and Shariah audit trails, often saves more time than a faster core would.

Front end, KYC and fraud

Most entrants build the mobile app themselves, sometimes on a digital banking platform or a backend-for-frontend layer. That's where the brand differentiates. Under BNM's e-KYC requirements, you onboard against MyKad and, for non-residents, passports, with liveness detection, document fraud checks and performance monitoring your board reviews. Options include global providers such as Jumio and Onfido and Malaysian specialists that tend to handle MyKad well. Scam prevention is no longer optional. Expect transaction monitoring from a vendor such as Feedzai, Featurespace, SAS or NICE Actimize, often paired with behavioral biometrics such as BioCatch, and connected to industry mule-account intelligence.

Payments connectivity through PayNet (DuitNow, IBG, FPX and JomPAY) and the RENTAS settlement link is on the critical path. Plan for certification cycles you don't control.

Indicative platform and compliance costs

The ranges below are typical 2026 market ranges for a foundational-phase digital bank targeting a few hundred thousand to a million customers. They vary widely by contract, scope, negotiating power and whether a shareholder brings existing technology. Treat them as planning brackets and confirm figures with vendors. Local pricing for Malaysian integrators and staff differs from global rate cards.

ComponentBuild / implementation (one-off, USD)Run cost (annual, USD)Main cost drivers
Core banking (licence or SaaS plus integration)4M – 15M1M – 4MProduct count, Islamic modules, account-based pricing tiers, SI rates
Digital front end (mobile app, web, BFF layer)2M – 7M1M – 3M (team)In-house vs agency, feature scope, accessibility and language support
e-KYC and identity verification200K – 600K0.40 – 2.50 per check, plus platform feeVolume tiers, liveness type, MyKad specialisation
Fraud, AML and scam prevention500K – 2M400K – 2MReal-time scoring, behavioral biometrics, sanctions screening
Cloud hosting and DR300K – 1M (landing zone)1M – 3.5MActive-active design, environments, data retention
Security operations and RMiT assurance500K – 1.5M700K – 2MManaged SOC, pen tests, independent pre-go-live assessment
Payments connectivity (PayNet, card issuing)500K – 2M300K – 1.5MCard processor choice, scheme fees, certification effort

Taken together, plan on USD 10 million to 30 million of technology spend before launch and USD 6 million to 18 million a year to run it. In-house engineering, risk and compliance salaries come on top. Here's a hypothetical worked example. A consortium with RM100 million of capital commits USD 18 million to pre-launch technology. That's roughly three-quarters of its minimum capital, leaving thin cover for losses, people and marketing, so shareholders would have to fund the gap. BNM looks closely at that funding plan. Put these costs in your application's financial projections, not in a separate IT budget.

The red flags in vendor pricing are consistent: per-account fees that escalate steeply once you leave the foundational phase, change-request rates that aren't capped, and assurance or audit support billed as optional extras when RMiT makes them mandatory.

A realistic timeline from application to public launch

For the first cohort, roughly ten months passed between the application deadline and the licence approvals, and around another 18 to 24 months before public launch. Plan on 30 to 40 months from serious preparation to a fully open launch. The table assumes BNM opens a window, or that you're joining an existing licensee. Treat the phase boundaries as planning estimates.

MonthsPhaseWhat has to happen
0 – 6Pre-applicationForm the consortium, confirm capital commitments, define the value proposition, appoint counsel, draft the target architecture and business plan with projections
6 – 16BNM assessmentSubmit the application; answer BNM queries, sit interviews and run fit-and-proper reviews; shortlist vendors in parallel without signing binding contracts
16 – 19Approval and mobilisationReceive approval, incorporate the licensed entity, appoint the board, CEO, CRO and CISO, sign core, cloud and SI contracts, complete cloud engagement with BNM
19 – 30Build and integrateConfigure the core, build the app, integrate e-KYC, fraud and PayNet; certify with payment networks; data migration is minimal but test data governance still applies
28 – 32Assurance and readinessIndependent RMiT assessment, penetration tests, DR drills, BNM operational readiness review
32 – 36Controlled launchPilot with staff and invited users, monitor fraud and incidents, report to BNM
36 – 40Public launchOpen onboarding, scale marketing within the asset cap

The build stage gets the attention, but the readiness review is where projects usually slip. If your RMiT evidence arrives late (DR test results, privileged access logs, vendor audit reports), expect the controlled launch to move back by a quarter or more. Allow at least eight weeks of buffer before any launch date you announce publicly.

Red flags that delay approval, and the decision rule for your platform

Applications and builds that struggle tend to fail in the same few ways:

  • Architecture designed for a different regulator. A stack built for a UK, EU or Singapore launch is a reasonable starting point, but it won't map line-for-line to RMiT's availability, cloud and assurance expectations. Re-baseline it against RMiT; don't assume equivalence.
  • Vendors with no Malaysian delivery presence. Remote-only vendors struggle with on-site audit requests, local certification and time-zone coverage for critical incidents.
  • Islamic products treated as a later phase. If Islamic financing is central to your proposition, retrofitting a conventional core usually costs more than choosing Shariah capability up front.
  • Financial projections that ignore run costs. BNM tests sustainability under the asset cap. Annual technology run costs of USD 10 million or more against a book capped near RM3 billion is a margin problem your business plan has to answer.
  • Weak scam controls. Given BNM's supervisory focus on fraud and mule accounts, a thin fraud stack is one of the fastest ways to draw supervisory concern.

The decision rule: shortlist only core and cloud partners that can hand you a completed RMiT control mapping, a reference deployment in a Malaysian or comparable Southeast Asian region, and a signed audit-rights clause before contract. Rank them by the cost of changing products during the foundational phase, not by feature breadth. Before you finalize the architecture or the application, take both to Malaysian regulatory counsel and request an informal engagement with BNM. Interpretations of the framework and of RMiT evolve, and only they can confirm how the rules apply to your structure.

Questions buyers ask

Can a new applicant apply for a digital bank licence in Malaysia right now?

Not through an open window, as far as we know. BNM ran one application window under the framework and indicated it would consider further licensing after reviewing the first cohort. New entrants usually wait for a future window, partner with an existing licensee, or invest in one. Check the current position with BNM and local counsel.

Does RMiT allow a digital bank to run its core banking system on public cloud?

Yes, subject to conditions. A bank must assess the risk against BNM's cloud expectations, secure board oversight and follow BNM's process for critical workloads. You still need to show resilience, key management, exit plans and audit rights. Malaysian cloud regions from the major hyperscalers make data residency easier to manage.

How much should we budget for technology to launch a digital bank in Malaysia?

Typical 2026 market ranges put pre-launch technology spend at about USD 10 million to 30 million, with run costs of USD 6 million to 18 million a year, before in-house salaries. Figures vary widely by contract and scope, so confirm them with vendors. Make sure the spend fits inside your capital plan, because BNM reviews financial sustainability.

How long does it take to go from application to public launch?

Plan on about 30 to 40 months. The first cohort waited roughly ten months for licence approval and then took another 18 to 24 months or so to launch publicly. The independent RMiT assessment and BNM's operational readiness review are the stages most likely to slip.

Written by the Vaultgridplus editorial team. Prices, features and rules change often - confirm details with the vendor or a licensed professional before you decide.