Punggol is already a live Singapore test environment for autonomous shuttle services. For a business or investor considering 2027–2037, the significant question is not simply whether self-driving vehicles can move passengers. It is whether an authorised mobility operator can provide safe, convenient transport at a cost that makes sense, and which properly qualified suppliers can support fleet maintenance, charging, remote operations, mapping, software, accessibility and passenger services. This is a guide to the business and industrial supply chain, not a passenger timetable or a promise that autonomous vehicles will replace conventional transport.

As of 1 November 2026, the Land Transport Authority has reported that Grab and ComfortDelGro operate Punggol shuttle services under a phased deployment approach. LTA’s 29 July 2026 update reported more than 11,500 unique riders as of 12 July and described plans for more flexible on-demand routing before revenue operations begin. On 7 October 2026, the Ministry of Transport stated that there were 15 autonomous vehicles in Punggol and that Grab intended to increase its fleet. On 19 October, LTA explained that current road deployments remain under the 2017 autonomous-vehicle sandbox, including operator insurance, safety and cybersecurity requirements, while a new legislative framework is being studied. These are dated official reports, not guarantees of 2037 fleet size, revenue or routes.

• Local anchor: Punggol’s authorised autonomous shuttle programme, with links to the wider Singapore mobility industry.
• Current participants: Grab and ComfortDelGro, working with relevant technology partners under LTA oversight.
• Possible business roles: Fleet support, approved maintenance, specialist mapping, operations software, charging, cleaning, accessibility and authorised testing.
• Regulatory boundary: Public-road deployments need LTA approval, safety assessment and compliant insurance; the public road is not an open testing ground.
• Commercial test: Real journeys completed safely, user satisfaction, effective route coverage and full operating costs.
• Important distinction: Passenger feedback, published trial counts and announced vehicle growth are not the same as profitable commercial demand.
• 2027–2037: Scenario planning for an emerging regulated service rather than an official vehicle or employment forecast.
• First action: Identify the genuine authorised operator and the precise supplier service it might need.

Define the autonomous mobility business correctly

An autonomous vehicle uses sensing, computing and control systems to perform some driving functions within an approved operating environment. In Singapore, the permitted deployment and the degree of human supervision depend on LTA’s assessment and conditions. An autonomous shuttle is a transport service, not merely a vehicle demonstration. It must integrate booking, access, route planning, maintenance, passenger information, safety response and customer support.

The business should be judged as a complete operating system. A supplier whose software performs well in a laboratory has not proved that it can support a public transport service. A vehicle capable of making a trip without direct steering intervention may still require safety supervision, remote support and carefully defined operating limits. The real commercial proposition is dependable transport for people who would otherwise spend more time or effort getting to their destinations.

What the Punggol programme has actually demonstrated

LTA announced three Punggol shuttle routes in September 2025 and began opening the services to public rides in April 2026. Its March 2026 release said Grab’s vehicles had accumulated more than 25,000 kilometres of autonomous driving by late February, while the July release reported substantial additional public participation. These figures show that trials progressed beyond a conceptual rendering. They do not establish the failure rate for every future vehicle, or the commercial economics of operations.

The July 2026 update also provided a useful market lesson: many riders preferred more direct routes and flexible pick-up points rather than a fixed loop. LTA said Grab was preparing an on-demand approach linking existing pick-up and drop-off points served by two routes. For founders, this illustrates why real customers matter. An impressive technical pilot may reveal that the service format needs changing before passengers are willing to pay.

A passenger guide is different from an industry guide

Our existing Punggol autonomous shuttle passenger guide and 2027 on-demand transport explainer address routes and what commuters can expect. The present article does not repeat instructions for booking a ride or invent a current fare. Instead, it examines which suppliers could potentially support an authorised operator and what makes a mobility business sustainable.

A route map tells a resident where a vehicle might take them. A business plan must also answer who owns the vehicle, what happens during downtime, how staff are rostered, who maintains the fleet, which insurance applies and how the operator pays its bills. These are different search intents and should remain separate. A founder who understands that distinction will be less likely to confuse public excitement with a viable revenue model.

The authorised operator sits at the centre of the value chain

A public autonomous shuttle service involves an operator responsible for meeting deployment conditions. Technology vendors may supply software, sensing systems or vehicles; maintenance organisations may handle approved service work; and various support firms may provide customer communication, charging or fleet facilities. The chain cannot be treated as a free marketplace in which any contractor may alter vehicles or run trials independently.

Before making a sales proposal, a supplier should identify the approved operator, purchasing authority and applicable technical arrangements. Does the operator procure a service directly, or does a technology partner already supply it under an integrated agreement? Does the contract permit outside maintenance? Who can approve changes to a safety-critical component? The answers determine whether a potential supplier has a real route into the market.

LTA’s safety assessment is not a formality

LTA’s autonomous-vehicle guidance describes safety assessment at the CETRAN test centre before vehicles can be deployed on public roads. Its framework includes different tracks for deployable and developmental solutions, with tests, demonstrations, documentation and operational-readiness requirements. A vehicle supplier should understand that demonstrating a robot in a private warehouse does not equal permission for public-road passenger transport.

The application must describe a realistic use case, intended roads or paths, vehicle capability and safety arrangements. LTA can impose conditions and monitor performance. A new company should not promise a public trial date without the required authorisation. The strongest commercial proposal often begins with a carefully bounded engineering or support service for an existing operator, rather than suggesting that the startup can independently launch an autonomous taxi network.

The October 2026 liability clarification matters

On 19 October 2026, LTA and MOT explained that autonomous vehicles on public roads were operating under the sandbox in the Road Traffic (Autonomous Motor Vehicles) Rules 2017. The current framework requires authorised operators to maintain liability insurance for third-party death, bodily injury and property damage arising from the vehicle’s use, including passengers. The agencies were also studying a new autonomous-vehicle Act.

These details affect contracts. A supplier should not claim that automation means there is no responsible human or entity. Nor should it assume that ordinary motor insurance automatically covers every autonomous use. Liability, technical responsibilities and reporting should be reviewed by appropriate professionals. As new legislation is being considered, a business planning years ahead must track legal changes instead of reusing the same 2026 assumptions indefinitely.

What counts as a commercially useful supplier service?

The most obvious supplier categories include vehicles and software, but an operating fleet needs much more. It requires approved servicing, tyre and component care, safe cleaning, charging, secure storage, user support, accessible information and dependable reporting. Some activities are ordinary commercial services; others are tightly linked to safety-critical systems and need specialised approval. The difference must be established before offering work.

For example, a firm that cleans approved vehicle interiors under an operator’s procedure is not the same as a company altering the vehicle’s steering or perception system. A software vendor providing aggregated service statistics is not automatically authorised to command vehicles. A firm may specialise in one legitimate support function and earn repeat business without claiming expertise in autonomous driving itself. Clarity of scope creates trust.

Mapping a route is much more than drawing a line

Autonomous vehicles need carefully validated information about their intended operating environment. LTA’s phased approach involves preparatory activities such as mapping and familiarisation before passenger service. Public roads contain crossings, buses, cyclists, delivery vehicles, pedestrians and temporary changes. A map can become outdated when roadworks occur or traffic arrangements change.

A mapping supplier should understand that the operating authority decides the validation process and approved scope. Data collection may involve permissions, safety procedures and privacy considerations. A commercial service might help keep authorised operational records current or organise changes, but cannot simply publish a volunteer-generated map as safety-certified. The buyer needs evidence that information is accurate, appropriately maintained and used only within its approved role.

Remote operations are a business function with strict boundaries

Autonomous mobility may involve remote oversight and assistance, depending on the vehicle and deployment approval. Remote support is not the same as leaving a vehicle unattended with a generic video call. Operators must satisfy LTA’s applicable requirements, which can involve systems, procedures, communications and assessments. A supplier supporting remote operations must know which decisions it is authorised to make.

Consider an illustrative vehicle that encounters an unexpected road obstruction. The operating procedure should define how the system responds, how the situation is escalated and when a human may intervene. The supplier should not claim that every problem can be resolved remotely. Safe procedures sometimes require stopping, waiting or changing the service plan. A company that respects these limits may appear less ambitious than one promising flawless autonomy, but it is more commercially credible.

Software integration and cybersecurity are operational necessities

A shuttle service may connect booking, vehicle state, location, customer communications, maintenance records and remote supervision. Each connection is a potential reliability or security dependency. A vendor should identify which systems it can access, what information it needs and the consequences if its software fails. A badly designed booking integration can inconvenience riders; an unauthorised control command could create much greater risk.

Security controls should include authorised accounts, appropriate logging, changes approved by the operator and a procedure for revoking access when the contract ends. The supplier must not test live vehicles or infrastructure without permission. LTA explicitly identifies cybersecurity among the requirements for AV operators. Commercial software can contribute value when it makes the service easier to manage without creating an unacceptable new dependency.

Why on-demand routing changes the operating economics

A fixed-route shuttle can be scheduled around a relatively predictable loop. An on-demand service changes the questions. Where are riders waiting? Can requests be grouped efficiently? How long is the vehicle idle? What happens when two customers need incompatible journeys? The operator must balance convenience with fleet utilisation, wait times and safe operating limits.

LTA’s July 2026 announcement described a planned move toward direct routing between existing pick-up and drop-off locations. The lesson is that passenger preferences and vehicle efficiency must be considered together. If every rider gets an individually direct trip, total empty movement could increase. If too many requests are grouped, journeys become inconvenient. A software supplier should evaluate trade-offs using authorised data and clearly explain which objective it is optimising.

One vehicle is not the same as one available vehicle

A fleet of ten vehicles does not necessarily provide ten vehicles for every scheduled hour. Some may need charging, servicing, cleaning, inspection or recovery after a fault. Demand varies across times of day and locations. Therefore, a business must estimate available vehicle-hours rather than multiply total fleet size by an imaginary perfect utilisation rate.

Consider a hypothetical small fleet of eight vehicles, each capable of ten scheduled service hours per day before downtime and other constraints. That represents a theoretical eighty vehicle-hours, but not necessarily eighty hours of passenger service. If maintenance and charging reduce availability by 15%, the result is sixty-eight vehicle-hours before further operational considerations. These are teaching figures, not published Punggol fleet statistics. They show why vehicle count alone is a weak business forecast.

Safety operators and human supervision remain commercially relevant

LTA has emphasised on-board safety operators during phases of autonomous trials. Different future deployment arrangements may have different approved supervision requirements. A company should not assume that the word autonomous allows all staff costs to disappear. Training, oversight, operational control and incident response remain part of the business.

Singapore’s Ministry of Transport has also announced measures to support drivers as the industry changes. Its July 2026 manpower package and October updates discuss conversion opportunities and the need to manage the transition. A legitimate mobility business should develop people rather than market automation as a shortcut around their livelihoods. The practical roles may include safety operations, fleet support, technical supervision and customer service. The exact staffing model must follow current regulations and actual operating conditions.

Why the October vehicle count should be read carefully

On 7 October 2026, MOT said that Punggol had 15 autonomous vehicles and that Grab planned to increase its fleet in subsequent months. That is a dated snapshot, not a number to project unchanged through the next decade. A forecast that simply doubles the fleet every year would have no basis in the statement.

Vehicle deployment depends on technical readiness, regulatory approval, actual demand, financing, depot capacity and the ability to support passengers safely. A larger fleet may improve availability in a dense market, but it also increases maintenance and operational cost. Entrepreneurs should distinguish vehicles announced, vehicles approved, vehicles deployed and rides actually completed. These are not interchangeable measures of commercial success.

What Waymo’s September announcement means for local suppliers

LTA announced on 18 September 2026 that Waymo would enter Singapore’s autonomous mobility landscape, with initial manual driving preparation expected in 2027 and a commercial ride-hailing ambition for 2028. Its intended initial location was not described as the existing Punggol shuttle network. Therefore, it would be misleading to present that announcement as confirmation that Waymo is operating Punggol shuttle routes.

The commercial implication is wider national interest in autonomous mobility, not guaranteed procurement at every estate. Suppliers must identify which operator and geography they are addressing. A company that builds a specialist service for one approved fleet might eventually serve others, subject to contracts and technical compatibility. But differing vehicles, operating frameworks and customer arrangements can require significant additional work. The right business plan acknowledges that complexity.

Fleet charging is a scheduling problem as well as an energy problem

Many autonomous shuttle platforms use electrical power, but vehicle specifications and charging arrangements vary. A commercial charging service needs to understand vehicle compatibility, electrical installation, safe operation and the time vehicles spend out of service. An apparently cheap electricity arrangement may still be poor value if vehicles cannot charge when they need to operate.

A supplier interested in charging infrastructure should identify the relevant owner and technical approvals. It must consider the location, power capacity, equipment maintenance and contingency if a charger fails. It should not install unapproved infrastructure at a public pick-up point. A workable operating plan might schedule charging away from expected demand peaks, but only if that suits the approved fleet and real usage profile. Good scheduling can matter as much as the hardware.

Accessibility is not an optional marketing benefit

Passenger transport must consider people with different mobility, sensory and communication needs. A service that requires complicated smartphone actions or provides unclear pick-up information may exclude some residents even if the vehicle drives well. A commercial operator should understand applicable accessibility requirements and use real customer feedback to improve the journey.

An accessibility supplier might offer clear information design, authorised user testing or appropriate service support. It must avoid advertising a vehicle as wheelchair-accessible without verifying the actual vehicle and operating arrangements. Older passengers, caregivers and people travelling with children can experience the same route differently. An operator who understands these experiences can make the service more useful and potentially improve retention. But accessibility should be evaluated through evidence, not by counting how many icons appear in an app.

Passenger safety feedback is valuable but not a profit-and-loss statement

LTA’s July 2026 release described positive safety feedback among survey respondents. That is encouraging for public acceptance, but it does not prove commercial viability or remove the need for ongoing safety monitoring. Survey samples, timing and question wording affect how feedback should be interpreted. Businesses should not rewrite a favourable response percentage as a prediction that almost every resident will buy a ride.

A service also needs information on repeat usage, avoided travel time, coverage of underserved journeys, rider complaints and cost per completed trip. Some of these measures may not be publicly available. A supplier should be transparent about the difference between official data and its own assumptions. Honest use of statistics is especially important when public trust and road safety are involved.

The true cost of a passenger journey

Fare revenue is only one component of a mobility business. Costs may include vehicles, depreciation or leasing, maintenance, software, network services, charging, staff, insurance, cleaning, depot facilities and customer support. Operators may also bear costs for mapping, testing and regulatory compliance. A founder cannot assume that eliminating conventional driving for one part of a journey eliminates all of these expenses.

Imagine a fictional service completing 200 paid rides a day at an average realised fare of S$4. That would represent S$800 daily revenue, before refunds or other adjustments. If daily operating expenses allocated to that service are S$1,100, it makes an operating loss despite vehicles moving passengers. These figures are illustrative, not an actual Punggol fare forecast or operator financial statement. The exercise forces the entrepreneur to consider full cost rather than celebrate ride counts.

Fleet utilisation can make growth economically difficult

Adding more vehicles can reduce customer waiting time, but each vehicle creates capital and support obligations. At low demand, extra capacity remains idle. At high demand, insufficient capacity leads to unacceptable waits. The operator must decide how much service reliability the market needs and what it is able to finance. Public transport objectives can also differ from private profit objectives.

A supplier proposing dispatch software should explain whether its system improves wait times, reduces empty driving or supports predictable operating costs. These aims may conflict. A model that focuses only on maximising vehicles in motion could produce long passenger waits or unnecessary travel. The commercial value lies in helping the operator manage trade-offs under its approved service policy, not in claiming to have solved transport with one algorithm.

B2B opportunities beyond the vehicle manufacturer

A small local company probably cannot design an entire autonomous vehicle platform, but it may have a relevant narrower service. It could supply appropriately authorised cleaning, depot support, documentation systems, approved equipment, accessible passenger information or operations analytics. Some services may be bought through the main operator; others may already be included in supplier agreements.

The firm should begin by mapping who currently delivers the function and what shortcomings are reported. If the operator has a comprehensive support contract, a new vendor may have no opportunity until a later procurement cycle. A supplier should avoid claiming affiliation without authorisation. Building a sound reputation with unrelated fleet or facilities customers may be a prudent step while the autonomous mobility market matures.

Mapping and testing services must preserve public safety

Testing technology on public roads is fundamentally different from evaluating a dashboard in an office. A developer interested in AV mapping or perception should use the authorised framework, suitable facilities and formal agreements. LTA’s published guidance explains its assessment process and how applicants should approach the authority. A founder should not drive a prototype autonomously on public roads merely to produce a demo video.

For non-road research, permitted controlled settings can offer valuable early evidence. A laboratory or simulation may help compare software versions, but it cannot fully prove public-road readiness. The company should document what each test establishes and which conditions are still unverified. The same caution applies to claims about safety kilometres: they need a clear operating context and reliable records.

Handling a service disruption responsibly

Suppose a shuttle stops because of a technical issue. The operator needs a defined procedure for keeping people safe, communicating with passengers and restoring or terminating the journey. A vendor may be responsible for a particular device or system, but should not independently intervene beyond its authorised role. The response might involve a safety operator, remote team, technical support or alternative travel arrangements, depending on the approved procedure.

An effective contract states who receives the fault report, who can make decisions and what evidence must be retained. A service that hides incidents for fear of negative publicity becomes less trustworthy. Carefully documented disruptions can help improve the product. Commercial readiness includes accepting that equipment sometimes fails and planning how to respond, rather than pretending the system will never need human help.

Business continuity is broader than having spare vehicles

A fleet can be disrupted by communications failures, severe weather, power issues, software defects or unexpected road restrictions. A spare vehicle helps only some of these problems. The operator should understand which services depend on the same network, cloud platform, depot and support personnel. A common dependency can affect several vehicles simultaneously.

A supplier offering continuity planning should identify critical functions, realistic recovery priorities and responsibilities across companies. It should use approved exercises to test communications and decision-making. The value is not an elaborate report that nobody reads; it is an operational plan that works when normal arrangements fail. A well-designed service may occasionally decide to suspend rides rather than operate outside safe conditions. That decision protects both users and long-term trust.

Procurement and vendor qualification

Autonomous transport combines public-facing service with regulated technology. A vendor may need to demonstrate financial stability, safety practices, cybersecurity controls, suitable insurance and credible experience. Some tenders will be bought centrally, while other contracts may involve vehicle manufacturers or operations partners. The right process depends on the actual buyer.

Prepare a capability statement that names the precise service and qualifications. Explain how the firm handles faults, support hours and confidentiality. If the company has never served an AV fleet, do not imply otherwise. It may have relevant experience in conventional fleet maintenance or secure logistics, but the differences should be recognised. A smaller, properly scoped contract can create genuine evidence for later expansion.

The workforce transition is part of responsible industry planning

Automation changes work, but not every change produces immediate unemployment. MOT’s July and October 2026 statements describe support for taxi and private-hire drivers interested in new roles, including relevant career-conversion programmes. The government has emphasised that AV adoption should not run ahead of support for affected drivers. This context matters for investors and employers making public claims.

An operator should consider the skills needed for supervision, remote operations, passenger assistance and fleet maintenance. Experienced drivers may have useful knowledge of routes, customer needs and safety practices. New roles can require additional technical training, and an employer should make realistic plans for that investment. A business that describes drivers as instantly redundant may misunderstand both policy and the operational value of experienced people.

2027–2029: a scenario of gradual service refinement

The near-term period could involve testing more flexible booking, learning actual customer demand and refining the operating model. It could also involve changes in regulatory rules as Singapore studies a more comprehensive AV framework. None of these possibilities should be treated as a guaranteed Punggol expansion timetable. A supplier should review the latest LTA notices before committing to costly premises or specialised equipment.

For a small company, the most practical strategy may be to prove one support function for an existing fleet customer. Document service quality, calculate full costs and establish authorised escalation. If the intended AV contract does not materialise, the capability should ideally still have value in a related legitimate market. Flexibility is more prudent than building the entire business around a guessed future vehicle count.

2030–2033: the market may reward repeatable operations

If commercial AV services become more established, operators may look for predictable fleet support and scalable systems. A supplier that has learned to manage documentation, maintenance schedules and safe handovers could become valuable. Yet technical platforms and operating rules may vary, limiting the extent to which one service can be copied to another fleet without further validation.

An expanding company must develop processes that do not rely entirely on one founder. It needs trained staff, tested incident procedures, version control and contractual clarity. A business that cannot handle a routine breakdown while its lead engineer is away is not ready to support a large service. The years 2030–2033 are planning checkpoints, not an official commitment that Punggol will operate a certain number of driverless vehicles.

2034–2037: evaluate resilience instead of predicting a robotaxi future

A decade can change vehicle technology, passenger expectations, energy cost and legislation. Some autonomous formats may prove valuable while others fail commercially. A supplier should avoid relying on one vehicle brand, proprietary interface or exclusive customer without understanding the risk. It may need to upgrade technical competence as new systems appear.

The more durable business advantage is the ability to support safe, dependable transport regardless of which technology supplier dominates. That means strong records, ethical incident handling, staff development and financial capacity to handle change. The 2037 endpoint is not a promise of fully autonomous transport everywhere. It is a reason to ask which organisational skills will remain useful if particular hardware or software becomes obsolete.

Three realistic small-business pathways

Pathway one: a conventional fleet-support SME

A workshop already provides lawful support for ordinary commercial vehicles. It wants to enter the AV market. The owner begins by identifying which conventional services might be transferable and which work is controlled by specialist manufacturers or operator approvals. It prepares a capability record, insurance review and a plan for authorised work only. It does not touch autonomous control equipment simply because it can service a normal vehicle.

Pathway two: an operations software supplier

A Punggol-based developer creates a simulation of booking demand and vehicle availability using fictional data. It demonstrates how the tool could make dispatch decisions more transparent and how staff could override an unrealistic recommendation. It does not claim to control live shuttles. If an operator expresses interest, the company works through formal security, integration and approval requirements before deploying anything.

Pathway three: an accessibility services company

A small design firm specialises in clear passenger information. It researches how different users understand pick-up points, delays and booking confirmation, working only with appropriate permission and privacy safeguards. Its proposed product might be better wayfinding or accessible instructions rather than driving technology. The firm tests whether confusion is reduced and documents limitations. This is a legitimate niche that can support an emerging mobility industry without claiming expertise in safety-critical vehicle controls.

Checklist for autonomous mobility suppliers

• Which authorised operator is the prospective customer?
• Is the service passenger-facing, fleet support, software, mapping or safety-critical equipment work?
• What LTA approvals and technical permissions apply?
• Who is responsible for liability insurance and incident escalation?
• Which systems or vehicle data may the supplier access?
• Are response times, staffing hours and equipment responsibilities clearly defined?
• What happens when a vehicle or connection fails?
• How will service quality and customer benefit be measured?
• How much cost is incurred when vehicles are idle or unavailable?
• Does the proposal assume a fare or vehicle count that has not been confirmed?
• What changes are being studied in the regulatory framework?
• Can the supplier perform without bypassing established safety procedures?
• How will existing workers be supported as operating requirements change?
• What is the exit plan if the contract ends or the technology platform changes?

A well-run supplier may answer that it is not yet qualified for a certain task. That is a useful commercial conclusion. A safety-sensitive transport industry needs partners that understand boundaries and can prove their work. A Punggol address alone will never be a substitute for that evidence.

Frequently asked questions

Are autonomous shuttles a real Punggol activity or only a proposal?

They are real deployed services under a phased programme. LTA published updates in March and July 2026 describing public rides and passenger feedback. Current routes, fares and booking arrangements should always be checked with the relevant operator.

Did LTA confirm permanent S$4 rides beginning in 2026?

The March announcement described an expectation for initial paid service, but July’s update revised the service approach toward on-demand flexibility before revenue operations. Do not treat the earlier projection as a verified current price.

Can a private business test self-driving vehicles on public roads without permission?

No. LTA’s assessment and approval framework governs trials and deployment. A company must use the authorised process and satisfy the applicable safety and insurance requirements.

Does a vehicle run completely without human oversight because it is autonomous?

Not necessarily. Singapore’s deployment approach includes defined supervision and operational requirements that vary by trial and approved stage. The exact arrangement should be verified for the service in question.

Is Waymo already operating Punggol shuttle routes?

The September 2026 announcement described planned preparations in another part of Singapore, not an operating Punggol Waymo service. It should not be presented as a confirmed local Punggol operator.

Will autonomous transport replace every local driver by 2037?

There is no reliable basis to make that claim. MOT describes phased deployment and support for affected workers. Technology, acceptance, safety and economic outcomes remain uncertain.

Your Next Step: identify a service the operator can actually buy

Choose one narrow problem in passenger communication, fleet maintenance, charging, operations data or another legitimate support category. Identify the approved operator and the decision maker who purchases that service. Determine the safety, licensing, data and insurance requirements. Build the smallest demonstration that can be evaluated without unauthorised access to vehicles or roads. Then calculate whether recurring revenue can pay for the people and equipment needed to keep the service reliable.

Official sources: LTA’s September 2025 Punggol deployment framework, LTA’s July 2026 service enhancement announcement, MOT’s October 2026 fleet and manpower statement, LTA’s 19 October 2026 liability and legislative update, LTA’s autonomous-vehicle assessment guide and LTA’s September 2026 Waymo announcement. Official positions can change; check them before taking a commercial or transport decision.