Grab (GRAB) Autonomous Vehicles: 9,000 Riders, No Fare Charged Yet

Grab Holdings (GRAB) put the first size figure on its autonomous vehicle service on the 2026-08-03 earnings call: the Ai.R / AIR shuttle in Singapore's Punggol district has served over 9,000 riders since January, and management set a date, the fourth quarter, to start charging a commercial fare [1]. On the 2025-02-19 call the company said only that it was in active discussions with regulators, and it named no partner [5]. In between, the line turned into a dated commercial service that still has not collected a fare.

Grab Holdings Limited (GRAB) is a Southeast Asian ride-hailing and delivery platform; riders order in its app, and those transport transactions produce its Mobility segment revenue. The AI here is the self-driving software in the vehicle, and Grab did not write it. It sources vehicles that can drive themselves from outside firms, WeRide, May Mobility and Momenta, then runs them as supply on its own ride-hailing marketplace. What Grab contributes is the platform layer: its own mapping and place data, its relationships with regulators and the local community, the Punggol depot where vehicles are parked and maintained, fleet dispatch and real-time alerting tools, and an AV Operations Command Center where staff watch the fleet centrally. The driver's seat becomes two new jobs: a safety operator who rides along and takes over when needed, and a remote operator who monitors several vehicles at once from that center. Both are existing Grab drivers retrained through the company's own Grab Academy. The application sits in Grab's core business, mobility, but its footprint is a pilot in one Singapore neighborhood.

How Grab's autonomous vehicle program evolved

It started as a concept. On 2025-02-19 management listed autonomous vehicles as a capital allocation direction and said it was in active discussions with regulators and pushing several partnerships forward [5]. It also noted itself that labor in Southeast Asia is cheap and road conditions and regulations differ, so adoption would take longer there. No vehicle was running, and no partner was named.

By 2025-04-30 there were signed agreements. Grab offered its fleet operating experience and its own general insurance license as what it brought to partners, but still declined to give a pilot timeline, because talks with various governments were under way [4].

On 2025-11-03 it committed capital and named the technology sources for the first time, grouping the WeRide and May Mobility partnerships and small minority equity investments under long-term bets [3]. It repeated on that call that unit economics parity between autonomous vehicles and human drivers is still a long way off, a limit not withdrawn since.

The turn came on 2026-02-11. Ai.R, built with WeRide, went live in Punggol, Singapore, having covered more than 25,000 kilometers with zero safety-critical incidents or near misses [2]. Drivers began training as safety operators, and management called Singapore the blueprint for the next three years of expansion. It gave no fleet size, no ride volume and no cost.

Where the Punggol AIR shuttle stands now

The operating team is in place and there is a revenue date for the first time. The scale is still very small.

On 2026-08-03 management said the AIR shuttle had served over 9,000 riders since January. Punggol now has a depot, more vehicles are being added this year, and through Grab Academy and work with the government it has certified more than 20 of its own drivers as safety operators, six of them further qualified as remote operators monitoring the fleet from the AV Operations Command Center. The service format changed too: instead of a fixed route, riders now set their own origin and destination in the app and are picked up point to point. That version opens to trial riders first, then to the public in the fourth quarter, when the company will begin charging a commercial fare [1].

The financial line this should eventually reach is Mobility segment revenue, and the path there is management's own: only when charging begins in the fourth quarter will the service produce fare revenue and enter that line. Until then it appears only on the spending side: the depot, the vehicles, the certification of safety and remote operators, the command center, and the minority investments previously grouped under long-term bets. Grab has disclosed no amount for any of those, and has not said whether they run through operating expense or capital expenditure.

The relationship therefore has to be recorded as not yet demonstrated. Every metric that survived verification is a rider count, a distance, a safety record, a staffing figure or a planning horizon; none carries a financial mapping. None of the more than 9,000 riders has paid a commercial fare, so that number cannot be read as demand validation or as a revenue run rate. It is measured on a different basis from the 2026-02-11 mileage figure, and the two time windows overlap, so the two cannot be added or divided into each other.

What to watch next

The operating judgment this history supports is that Grab's AV line is still entirely on the cost side of the financials, and that its cost structure runs against the intuition that AI replaces labor: retraining drivers into safety operators and remote operators adds a layer of human supervision on top of the fleet. On 2026-02-11 management itself positioned autonomous vehicles as a reliability buffer for supply rather than as a replacement for drivers.

The ceiling is management's own as well: more than half of the transactions in the region are two-wheel rides priced under one dollar, where autonomous vehicles are uneconomic, and Singapore accounts for only a small part of the region's four-wheel transactions. This line will not drive group margins within any foreseeable period. What it is really buying is operating experience and a regulatory position in one small permitted market. The first checkpoint verifiable against a financial report is the first commercial fare in the fourth quarter of 2026. The fare level and the size of that first fare will rewrite the judgment.

Application assessment

• Autonomous vehicles on Grab's mobility platform - a partner-supplied AV stack operated by Grab as a hybrid human-plus-autonomous fleet. | Business position: core operations | Adoption stage: pilot | Scope: limited scope | Value type: revenue growth

Sources

[1] Drillr - Grab Holdings Limited (GRAB) - 2026-08-03 - earnings call

"Since January, our AIR shuttle has served over 9,000 riders."

[2] Drillr - Grab Holdings Limited (GRAB) - 2026-02-11 - earnings call

"Ai.R has covered over 25,000 kilometers with zero safety critical incidents or near misses. This is the highest mileage recorded and the most data collected by any AV operator in Southeast Asia. Our in-house fleet operations tooling will also enable real-time alerting for AV issues, which allows our operations center to respond quickly to potential incidents."

[3] Drillr - Grab Holdings Limited (GRAB) - 2025-11-03 - earnings call

"You've seen the announcement that we made with WeRide for an example, May Mobility as part of our strategic pillar in terms of making sure that we are the pioneer and we're leaning in, in terms of autonomous vehicles deployment here in Southeast Asia."

[4] Drillr - Grab Holdings Limited (GRAB) - 2025-04-30 - earnings call

"Our intention is to really be at the forefront of the exploratory use of this new technology. I won't give you a timeline for pilots because we obviously are in discussion with various governments to get that going."

[5] Drillr - Grab Holdings Limited (GRAB) - 2025-02-19 - earnings call

"So going forward, I can confirm we are in active discussions with regulators. We intend to work closely every government in Southeast Asia to drive this forward."