Singapore Will Pay Up to 70% of Quantum-Safe Migration Pilots, Critical Infrastructure First

October 14, 2026 – The Cyber Security Agency of Singapore (CSA) will pay up to 70% of the cost of pilot projects that test quantum-safe migration in real operating environments, with priority for the critical infrastructure owners it has asked to finish migrating by the end of 2031. Josephine Teo, Minister for Digital Development and Information, announced the Quantum-Safe Kickstarter Grant in her opening address on the second day of Singapore International Cyber Week.

Teo also announced an advisory for organisations outside critical infrastructure that are exposed to quantum risk. It recommends that they have a migration plan by the end of 2027 and complete migration by 2035. In her address she said organisations “do not become quantum-safe overnight, just because they want to.”

The grant page limits eligibility to non-government organisations registered and incorporated in Singapore, although CSA’s press release describes the grant as open to all organisations. CSA said it will give priority to owners of Critical Information Infrastructure (CII), the systems designated under Singapore’s Cybersecurity Act in sectors that include banking and finance, energy and info-communications. CSA’s grant page says attackers are already expected to be stealing encrypted data in order to decrypt it once quantum computers mature.

Applications are open now, and the grant period runs from November 1, 2026 to October 31, 2027. Each project must be completed within six months of its award and by October 31, 2027, though CSA may extend a project case by case.

Applicants must contribute at least 30% of project costs in cash. Internal staff time does not count toward that share, and GST is excluded from the grant and borne by the applicant. CSA aims to decide within four weeks of receiving a complete application and awards grants on a first-come-first-served basis, subject to its assessment of scope and strategic value.

Organisations can run a pilot in-house or with a solution provider, including a vendor based overseas, as long as the solution is deployed and evaluated in a Singapore environment. CSA can share a non-exhaustive list of vendors and said inclusion on it is not an endorsement. Neither the grant’s info pack nor CSA’s press release discloses a total budget or a per-project cap.

For pilots run with a solution provider, CSA pays half the grant once the project is approved, the letter of award is signed and the commercial agreement is submitted. It pays the other half when it accepts the final report. In-house pilots receive no upfront payment and are paid in full after completion, report acceptance and submission of cost documentation.

CSA listed five illustrative priority use cases:

• end-to-end migration of one bounded, low-complexity application or system, such as secure web browsing, VPN, identity verification or certificate management
• cryptographic discovery and inventory across systems, applications, certificates, protocols, devices, hardware modules and supply chains
• assessment, planning or migration of operational technology (OT), embedded devices, network appliances and other legacy systems with long lifecycles
• crypto-agility implementation
• limited trials of quantum key distribution (QKD) or other quantum-enabled security, complementing post-quantum cryptography (PQC) without necessarily replacing it

Every funded pilot ends with a report in eight sections, covering what did not work, the limitations observed and practical difficulties such as vendor responsiveness. The eighth section asks for the approximate time and resources the organisation and its provider invested, which CSA said will help it calibrate future guidance. CSA said it may publish aggregated or anonymised findings across pilots and will not publish an individual organisation’s report without its explicit consent.

CII owners remain on the timeline CSA issued to them in June. They will need to submit quantum-safe migration plans for their CII to CSA by March 31, 2027. Migration of CII systems must be complete by December 31, 2031. CII systems contracted from January 1, 2028 must either support quantum-safe algorithms or be capable of supporting them by the end of 2031, according to CSA’s guidance for CII owners.

The guidance does not require a complete roadmap in the first plan. Owners must tell CSA why any information is missing, update their plans as migration proceeds and flag material deviations, and can ask in advance for a later submission date, subject to CSA’s approval.

The new advisory said Singapore’s timelines are aligned with international guidance, in which 2031 is a common target for critical systems and 2035 for everything else.

CSA and the Government Technology Agency (GovTech) are separately running a request for information on the readiness of the quantum-safe product and services market. Responses close on November 27, 2026, and require an account on GeBIZ, Singapore’s government procurement portal.

My Analysis

In July I wrote that CSA, by setting 2031 as the completion date for CII, was asking owners for a harder transition than the SHA-1 retirement in less time, and that I didn’t expect every operator to make it. I still think both. On Wednesday CSA, the agency that set the date, offered to pay for part of the work needed to meet it. Every regulator that has set a PQC deadline for privately owned infrastructure should do something similar.

PQC migration is the largest cryptographic change most organisations will ever make. One program plan I worked on contained more than 120,000 discrete tasks. In the illustrative breakdown I later published for a program of that size, roughly 30,000 tasks were direct remediation. The other 90,000 or so were the work needed to organise, support and verify that remediation, such as discovery tooling, governance, vendor contracts, test environments and training. A regulator that publishes a completion date and stops there leaves each operator to work out the cost, the method and the supply chain alone.

I have argued that the deadlines regulators set matter more to a CISO than Q-Day predictions, because boards fund obligations. A CII owner reading CSA’s June guidance learns when it has to finish. It still has to work out for itself what migrating an HSM-backed payment switch will cost and how long a SCADA vendor needs to ship firmware signed with ML-DSA (formerly CRYSTALS-Dilithium). It also has to guess whether vendors will have compliant products, or at least contractual upgrade paths, in time for the 2028 procurement milestone. Governments can help operators answer those questions in several ways. Money is the most direct, and Singapore is now using it.

The economic case for paying is as strong for a large bank as for a small firm. The first operator to migrate a payment switch or a SCADA network learns lessons that its competitors, suppliers and regulator all benefit from, and it pays the whole cost of learning them. By covering part of that cost in exchange for a report, CSA buys those lessons for everyone else.

Korea, the EU and Singapore’s central bank funded PQC pilots before CSA

CSA is not the first government agency to pay for PQC migration pilots. South Korea has funded PQC “pilot transitions” on national infrastructure since 2025. The Ministry of Science and ICT (MSIT) and the Korea Internet & Security Agency (KISA) run the program under a 2023 master plan that targets migration of the national cryptographic system by 2035. The 2025 round covered energy, healthcare and public administration. This year KISA expanded the program to telecommunications, finance, transport, defense and space, with up to ₩900 million in government funding per selected consortium and ₩4.5 billion in total for 2026.

KISA selects one consortium per sector, and the infrastructure owner may lead it. The two 2026 selections I could trace are both vendor-led. One consortium is migrating the card-payment infrastructure of the card issuer Hana Card, and another is applying PQC to KREONET, Korea’s national research network. Under KISA’s 2026 call booklet, the government pays up to 50% of project costs for large and public enterprises, 70% for mid-sized firms, 75% for SMEs and 100% for non-profits. Korea gets a few deep reference cases a year, three in 2025 and five in 2026, each selected by KISA.

Singapore’s central bank moved first at home. In July 2024 the Monetary Authority of Singapore (MAS) added a quantum track to its Financial Sector Technology and Innovation (FSTI 3.0) grant scheme. The track included a security grant covering up to 30% of pilots exploring PQC and QKD. Only financial institutions qualified, and FSTI 3.0 ran until March 2026. MAS announced its successor, FSTI 4.0, on August 31, 2026, and named quantum technology as one of its focus areas, but I could not find a published funding rate for quantum security pilots under it.

MAS launched the 2024 grant two years before its managing director, Chia Der Jiun, said in July 2026 that MAS will issue supervisory expectations for financial institutions, adding that “our aim is for FIs to achieve quantum resilience before the end of this decade.” A Singapore bank that owns designated CII therefore answers to two dates, MAS’s end-of-decade aim and CSA’s December 2031 deadline.

The European Union funds consortia through its Digital Europe Programme. A 2024 call for PQC deployment in industrial sectors offered a €22.25 million budget, at a 50% funding rate, for large-scale PQC demonstrators in automotive, automation, finance, control systems and energy, with an indicative EU contribution of €5–7 million per project. The call closed in March 2024, more than a year before the EU published its coordinated roadmap with 2030 and 2035 milestones, and it was not framed as funding to meet any deadline.

Washington budgets for its own systems. The White House Office of Management and Budget estimated in July 2024 that moving prioritised federal civilian systems would cost about $7.1 billion between 2025 and 2035. I think that figure is low, because agencies produced their estimates before finishing their cryptographic inventories. A report released on October 6 by the Government Accountability Office, the US Congress’s audit agency, found only one of 24 agencies with a complete inventory of its priority systems. I found no broadly available federal grant for private operators comparable to CSA’s.

Korea is the closest precedent. MSIT owns the national 2035 target and also pays for pilots, and a Korean infrastructure owner can lead its own consortium. But Korea runs one competition a year and funds one project per sector, while CSA takes applications from any eligible organisation on a first-come-first-served basis for as long as the grant period leaves time to finish a project. CSA’s 70% ceiling applies to every applicant, where Korea’s ranges from 50% for large and public enterprises to 100% for non-profits. And CSA set its 2031 date for a defined group of designated operators, so the organisations most likely to apply are the same ones that must file plans in March. As far as I can find, no other regulator has set migration deadlines for a designated group of critical infrastructure operators and then offered co-funding to those same operators.

CSA’s pilot reports could supply the cost data its Handbook lacks

To my mind, the most useful part of the package is section eight of the mandatory pilot report. It asks each funded organisation and its provider for the approximate time and resources they spent, so that CSA can “calibrate future guidance on what projects realistically require.”

My main complaint about CSA’s Quantum-Safe Handbook in July was that it gave no cost benchmarks at all, whether as a share of security budget, a per-system range or a headcount ratio. I said CSA would be better placed than almost any regulator to collect them once migration plans arrived. With the grant, CSA will get reported effort from pilots on real systems in Singapore, including pilots that go badly, and that is a better source than plans. Korea’s KISA also collects final reports and audited cost settlements from its pilots, but CSA asks for effort explicitly and says it will use the answers to calibrate its guidance.

The pilot data could also prove me wrong. In May I published an illustrative cost model that puts a large enterprise’s program at roughly $16–70 million. I built it on assumptions because nobody had better numbers, and I’d rather see real figures replace them than keep defending them.

CSA will get usable benchmarks from section eight only if every pilot reports effort the same way. The template asks only for approximate time and resources, so a pilot that kept two engineers busy for six months could look identical to one that spent five of those months waiting on a vendor. CSA should publish a common reporting method before the first pilots start, one that separates:

• staff effort from elapsed time, including time spent waiting on vendors
• internal labour from external fees, licences and replacement hardware
• first-time setup from repeat deployment
• completed migrations from discovery work, feasibility studies and inconclusive trials

CSA also asks for the failures. Applicants must report compatibility problems, performance constraints, maturity gaps in standards and trouble with vendor responsiveness, and CSA says inconclusive results are as welcome as clean ones. Vendors publish success stories. Operators rarely publish how many weeks they lost waiting for an HSM firmware update, and in my experience the next operator to attempt the same migration waits just as long.

Washington dealt with the same reluctance to share bad news during Y2K. On October 19, 1998, President Clinton signed the Year 2000 Information and Readiness Disclosure Act, which limited, within conditions, the liability of companies that shared readiness information. It also gave them a temporary antitrust exemption to coordinate fixes, with exclusions for price fixing, market allocation and boycotts. Companies could then tell each other what they had found with less fear of being sued for it.

By promising not to publish an individual report without consent, CSA protects operators from public exposure. But the agency reading a pilot report is also the regulator that will assess the same operator’s migration plan and its progress toward 2031. Operators need to know how CSA will keep what it learns from pilots separate from its supervisory assessment, and they will write section eight more candidly once CSA tells them.

Four gaps in the Kickstarter Grant’s design

Pilot scale. CSA lists end-to-end migration of one bounded, low-complexity system as its first use case. That suits a first project, but an operator learns little from it about what a full program costs. A team migrating a single TLS endpoint or VPN concentrator sees the cryptographic engineering in detail.

The same team sees far less of the work around remediation that makes up most of a large program, such as building the inventory pipeline, renegotiating vendor contracts, standing up test environments and running governance across workstreams. CSA’s list includes discovery and OT pilots too, but nobody can scale the effort of a six-month project directly into an organisation-wide cost. CSA should ask each pilot to estimate separately what rolling its approach out across the organisation would take. Without that estimate, the benchmarks CSA eventually publishes could understate program cost by a wide margin.

Timing against the March 2027 plans. CII owners must submit their migration plans by March 31, 2027, 150 days after the grant period opens on November 1. CSA aims to decide on an application within four weeks, and a pilot can run for up to six months. A six-month pilot starting on November 1 would run until the end of April 2027, a month after the plan is due.

CSA has already allowed for this in its guidance for CII owners. It accepts an incomplete first plan if the owner explains the gaps, expects plans to be updated, and considers extension requests made in advance. It hasn’t said when pilot findings should feed into those updates, or whether it will release interim lessons before the grant period ends in October 2027. Owners filing in March would plan better with both answers.

An undisclosed budget. With no published envelope and no per-project cap, applicants cannot tell how many pilots CSA will fund. Under first-come-first-served, the organisations that can apply fastest are the ones with PQC teams already in place, which in my experience means the large banks and telcos.

Smaller organisations also have to pre-finance the work. An in-house pilot is paid nothing until it is finished and its report accepted, and a provider-led pilot gets half the grant up front. In both cases a smaller organisation has to finance much of the project for months on top of its own 30% share, and a mid-sized logistics firm covered by the 2035 advisory may decide it cannot carry that for six months. If CSA published an envelope, even an approximate one, smaller organisations could at least plan an application instead of guessing.

The vendor list. CSA says inclusion on its vendor list is not an endorsement and that listed and unlisted vendors are evaluated on the same basis. The paperwork is not equal, though. An applicant that chooses an unlisted vendor has to supply additional information about it, and one that chooses from the list doesn’t. CSA could close that gap by publishing how vendors get onto the list and how often it is updated.

What other regulators can copy from Singapore

In most of the jurisdictions I follow in my Global PQC Migration Timelines deep dive, a cyber agency publishes one or more dates and leaves privately owned critical infrastructure to fund the work from its own budget. I think that is the wrong default. Regulators that set a 2030 or 2031 completion date for private operators and offer nothing beyond guidance documents are asking those operators to carry a national-security cost on their own balance sheets, and they should expect thin plans in return.

Regulators elsewhere can copy five of Singapore’s 2026 measures, and only the first needs a grant budget:

1. Fund the hard cases first. CSA’s priority list names OT, embedded devices and legacy systems with long lifecycles, where operators have the fewest options and where OT migrations take longest.
2. Collect effort data and publish it. Asking for effort in the pilot report costs CSA little. Turning the answers into published ranges will take analysis, but every operator in the region could then use them to defend a budget.
3. Map the supply side. CSA and GovTech are using their request for information to ask vendors and integrators what they can deliver today. In most programs I have seen, vendor readiness is the longest item on the critical path.
4. Use procurement to move vendors. CSA does this for designated operators through its 2028 milestone, under which it will accept CII systems contracted from 2028 that do not yet support quantum-safe algorithms, provided they can by the end of 2031. Operators will therefore be buying vendors’ upgrade commitments as much as products, and they need to write them into their contracts. The Canadian Centre for Cyber Security published recommended PQC contract clauses for federal procurement, ITSM.00.501, in September 2025.
5. Make it safe to report what went wrong, as the Y2K disclosure act did and as CSA’s anonymised reporting partly does.

A regulator without a grant budget can still run a voluntary pilot-reporting scheme, publish effort ranges and survey its vendor market. Regulators with a budget should study Singapore’s terms closely. Seventy percent of a six-month pilot is a small price for the evidence a regulator needs to judge whether its own deadline is achievable.

Next steps for CII owners, other organisations and vendors

CII owners. Apply this month for a discovery pilot scoped to your crown-jewel systems. A plan written without discovery can’t be checked against anything, and CSA is now offering to pay most of the cost of doing it. A pilot that CSA approves quickly and that you keep to four months will report before the March 31 plan is due, so start your internal approval now if it usually takes longer than CSA’s four-week review target. A second, clearly differentiated application from your group, for example from a subsidiary, should be an OT or legacy pilot, because that is where the 2031 date is least credible and where CSA has asked most plainly for evidence. CSA assesses each application separately and may limit the number of awards to one corporate group.

Other exposed organisations in Singapore. The grant is open to you if you’re a non-government organisation incorporated in Singapore, though CII owners come first. The advisory’s end-2027 planning date gives you nine months more than CII owners have, so you can run a funded pilot in 2027 and write the plan from its results.

Vendors. The RFI closes on November 27 and is the obvious place to tell CSA and GovTech what you can deliver today. Expect the pilot reports to describe your products’ compatibility gaps and response times, and assume CSA will read every one of them, anonymised or not.

Whichever group you’re in, write section eight honestly. An operator that under-reports effort to look efficient helps produce benchmarks that make its own next budget request harder to defend, and every other operator’s too.

When the grant period closes in October 2027, CSA will hold reported effort data from every pilot it funded. I’ll be asking CSA to publish it in a form other operators can compare.

From the author