Everyone, and welcome to Grab's third quarter 2026 earnings call. I am Douglas Eu, Director of Strategic Finance and Investor Relations at Grab. Joining me today are Anthony Tan, Chief Executive Officer, Alex Hungate, President and Chief Operating Officer, and Peter Oey, Chief Financial Officer. During this call, we will be making forward-looking statements regarding future events, including our business and financial performance. These statements are based on our current beliefs and expectations. Actual results could differ materially due to a number of risks and uncertainties as described in this earnings call, in the earnings release, and in our Form 20-F and other filings with the SEC. We do not undertake any duty to update any forward-looking statements. We will also be discussing non-IFRS financial measures on this call, and these measures supplement but do not replace IFRS financial measures.
Please refer to the earnings materials for a reconciliation of non-IFRS to IFRS financial measures. For more information, please refer to our earnings press release remarks and supplemental presentation available on our IR website. For today's call, Anthony will deliver opening remarks, after which we will open the floor for questions. Just a reminder, we accept questions via our IR mailbox at investor.relations@grab.com. Please do submit your questions ahead of time, and we will add them to the Q&A queue. With that, I will hand it over to Anthony.
Thanks so much, Doug. Good day, everyone, and thank you for joining us. The third quarter of 2026 was a quarter of several milestones. For the first time, all three of our core segments are profitable, with financial services achieving segment adjusted EBITDA profitability, including the contributions from Stash and Superbank. Adjusted EBITDA reached a record $237 million, up 73% year-on-year, our 19th consecutive quarter of adjusted EBITDA expansion. On-demand GMV grew 22% on a constant currency basis, and monthly transacting users reached a record 57 million. With a clear line of sight to the fourth quarter, we are raising our full-year guidance on both group revenue and adjusted EBITDA for the second time this year. We also reiterate our 2028 expectations of a group revenue compounded annual growth rate of over 30% and adjusted EBITDA of $1.7 billion.
These expectations include the proposed acquisition of Atome, which remains subject to completion. We delivered this in a challenging environment. Elevated fuel prices continued to pressure driver operating costs, persistent inflation generated or tightened household spending, and Indonesia's new two-wheeler commission framework took effect in July. We leveraged our ecosystem advantages to grow revenue while expanding profitability, and our mobility and deliveries margins both expanded quarter-on-quarter to 8.9% and 2.7% respectively. This is what we mean when we say our ecosystem compounds with more users, more services per user, and more of each dollar reaching the bottom line. That strength is also what allows us to stand with our driver partners. We know elevated fuel costs have weighed directly on their take-home earnings, and we continue to support them through this period.
Beyond immediate relief, we are also equipping partners with tools that help them earn more on every hour they spend on Grab. AI now powers every layer of the Grab ecosystem, and we are building AI-F irst with Heart to simplify experiences for our users and drive tangible earnings growth for our partners. As AI agents become a new way for consumers to transact, every order they place still needs to be fulfilled on the ground through the drivers, the merchants, and logistics, and licenses we have built for over more than a decade. With over 20 billion cumulative transactions on our platform, our models have a deep real-world understanding of how Southeast Asia moves, eats, shops, and borrows. We are doing this while structurally lowering our compute costs, with costs per AI interaction down 42% quarter-on-quarter as AI interactions grew 60%.
In September, we also announced a partnership with OpenAI to equip 30,000 driver and merchant partners with practical AI skills over the next two years, starting in Singapore and then Thailand, and expanding to Indonesia and the Philippines later this year, followed by Malaysia and Vietnam in 2027. Southeast Asia has been our home for nearly 15 years, and our conviction in the region has never been stronger. It is one of the world's fastest-growing economies and is projected to expand well above the global average in 2026, according to the latest IMF forecasts. With all three core segments now profitable and our outlook raised for the second time this year, Grab has more engines contributing to profit than at any point in our history. I share that conviction personally, and I have purchased additional shares in Grab.
Our mission to drive Southeast Asia forward by empowering millions across the region remains central to our strategy, and every capability we build serves our triple bottom line of delivering long-term shareholder value, creating meaningful impact for the communities and partners in our ecosystem, and operating responsibly across Southeast Asia. Thank you so much, and let us open it up for questions.
Great. Thank you, Anthony. We will now begin the Q&A session, and we encourage you to submit questions throughout the webinar via email. We received some questions coming in. The first question is on guidance upgrade, and this is a question for Peter. The question comes from several analysts, Ranjan, Divya at Morgan Stanley, Piyush at HSBC, Jiong at Barclays, and Wei from Mizuho. The question is, "Pleased to see the guidance upgrade, but why not raise the 2026 EBITDA guidance higher, as the current guidance indicates that the fourth quarter EBITDA may be stable relative to the third quarter?" Question then goes on to ask, "What are you assuming for fuel support, FX, and the festive season investments for the fourth quarter?" This question is for Peter.
Yeah, sure. That's a great question to open up. When you look at our guidance revision today, what are we at? We're at $790 million - $800 million. This is from our previous range, just to reset, from $720 million to $740 million. That's a roughly about a 58% to a 60% year-over-year expansion on our EBITDA. In context, we are now operating at the highest margins level in Grab's history. If you look at the year-over-year, it was a 560 basis points improvements in terms of EBITDA margin that we are continuing to demonstrate as our business continues to perform. The progression from Q3 to Q4, which is what the heart of the questions, there are a couple of deliberate strategic choices that we as management team, going into Q4, feel that it's really important.
If you look at fourth quarter, fourth quarter is a high-return investment window. It's one of our busiest season for Southeast Asia itself. It's the festive season and a very high volume tourist travel also at the same time. We saw some snippets of that in October with a lot of the events and also coming up also in December itself. Anthony just spoke about driver supply and some of the challenges that we've got here also. The elevated fuel prices has continued. We saw it in the second quarter, we saw it in the third quarter, and we're expecting the same in the fourth quarter. We need more drivers on the platform. If you look at the third quarter, our active driver base was the highest ever on the Grab ecosystem itself.
We increased driver base by 20% on a year-over-year basis, and we need more drivers out there. Securing driver partner capacity is really critical as for us going into the Q4. So we're leaning in Q4, to make sure our driver community continues to be strong and also to help our passengers also to have access to rides reliably as well as affordable at the same time. We're also deliberately deploying targeted marketplace investments across the board, both on the driver side also, and to also having making sure they have access to vehicles at the same time. Because if we fail to have drivers on the platform, it will just be leading to higher prices, higher consumer waiting time, which is not a great customer experience. That's one part which is really important, how we think about Q4, especially around the mobility.
The second part, going into this very busy season is also supporting the high growth product launches. If you're reading the earnings remarks today, what did we launch? We launched GrabMall in the Q3, and we're also rolling out several digital banking initiatives, the GXS Credit Card in Singapore, and also in Indonesia, we launched in conjunction with our collaboration with Atome, OVO Pinjaman, the lending product for our consumers with Kredit Pintar. These are very exciting both delivery products as well as financial services products. So we're allocating marketing and also leaning in to make sure these products continue to pick up the traction in Q4 and also as we get into 2027. On the cost structure itself, we see that regional corporate costs to be fairly stable in the second half to the first half.
That is really the package of how we think about the Q4 itself. My guidance philosophy has always been to be on the prudent side and also be transparent with our shareholders, but also be grounded in what we are seeing on the ground. We are seeing some of the challenges on our driver community that we talked about earlier. We are setting expectations to ensure that we can reinvest with agility, but also capture the seasonal market share. We want to finish the fourth quarter strong, and we are on track to do so profitably at the same time, also deliver a great customer experience for our customers, also for making sure our drivers and our merchants also are also increasing their earnings at the same time.
Thank you, Peter. We received this next question set, came from several analysts. It is all pertaining to the deliveries segment. We have segmented them into three parts. The questions came from several analysts like Divya, Morgan Stanley, Alicia, Citi, Piyush of HSBC, Ranjan of JPMorgan, and Thomas, Jefferies as well. There is a question pack for Alex. The three-part questions, the first is on margin, second is on ads, and then third is on clarification on GrabMall. I will read basically each of the three components. The first component is on the delivery segment. Is the 2.7% EBITDA margin a new structural baseline? What drove the outperformance, and how much of this was enabled by AI-driven unit cost efficiencies? Should we expect the 2.7% segment margin to be a new base? The second part of the question is then on advertising.
Can you update us on advertising penetration relative to deliveries GMV, and what outperformed in the quarter? The third part is on clarification on GrabMall. Could you expand more on the GrabMall strategy? This is a question pack for Alex.
Well, thank you all. I guess this is the most popular question so far, so let me try and go a bit detailed in those three parts. You are right, deliveries margin performed very strongly, 2.7%, up from 2.1% in the prior year, so very strong in the quarter. That does reflect primarily the operating leverage that we are getting from the scale because deliveries is growing fast, so 21% GMV growth year-on-year. As Peter said earlier, our control of the corporate cost, the regional corporate cost has been good, so pretty much flat in the second half versus the first half.
That gives us strong operating leverage, enhanced by very widespread deployment of AI, both in terms of the productivity that we as Grabbers benefit from, but more importantly, if you think about our central purpose. It is driven by continued improvements in the performance of the marketplace that we run, increasingly using AI to do the optimization that we need to make sure we get the best combination of reliability of service and affordability for our users. That is going very strongly. The other driver of the improvement in margin is ads. Ads are growing strongly, as our scale attracts more of the endemic advertisers who sell food on the platform, and increasingly also brings in new FMCG advertisers who start to get attracted by that growing reach, and are now very engaged also in partnering with us to grow GrabMart and GrabMall.
We have been able to do this at the same time as we increase driver incentives. As Peter mentioned, we are leaning in to help drivers during these difficult times, so you do see increased incentives both for mobility and for deliveries for drivers. The operating leverage is even bigger than the increase in incentives that we have given out to drivers. That shows the strength and the overall reach that we are getting now with that scale. We still see plenty of opportunities to invest, though, in the deliveries segment, using product-driven growth, which has been a key factor in our success over the last couple of years. We are expecting, as you remember from the last quarter, to invest in product growth in the GrabMall marketplace area.
Something like $160 million we set aside in the 2028 adjusted EBITDA guidance to fund those investments, driving better affordability in groceries and driving further growth in both GrabMall and GrabMart. But in spite of these investments, we expect deliveries margins to improve progressively year by year as we make progress towards our longer-term margin target of 4% +, which you will recall we have been guiding towards over the last couple of years. Then the last part of the strategy, I believe, was the GrabMall question, was the GrabMall strategy, so let me talk about that. We launched GrabMall this quarter. It is a branded retail commerce channel within the deliveries ecosystem. It has been inspired by the FMCG advertisers who have been working with us on advertising. They are now keen to partner to do closed-loop endemic advertising, where we actually fulfill the transaction as well.
Because of our very strong last-mile quick commerce capabilities, we are able to do that probably better than any other partner in the region. It is a good combination. We tap on our ecosystem advantage, and then we have that very large high-frequency user base and the proven fulfillment network. Then we also know how to work well with merchants in terms of making sure that we tap into understanding their inventory so we can get very good online fulfillment rates for customers as well. The additional benefit here, once Atome is approved, is of course that we will be able to then have better monetization capabilities via lending as well on this new base. We will get not just the advertising improvement in margin as we grow GrabMall and GrabMart, but we will start to get stronger lending monetization as well with those highly developed Atome consumer lending models.
We will continue to pilot GrabMall, which we just launched across select markets, and we are seeing good uplifts right away from these pilots, and we will start to expand that across the region.
Thank you, Alex. The next question is on AI, and specifically on AI proof points. Question came from Alicia, Citi and John of Daiwa. This is a question for Anthony. Anthony, could you elaborate on how AI agents and GenAI features are currently driving user engagement and transaction velocity across the platform? Specifically, what early trends are you observing in terms of cross-sell, in-app time spent, and basket size uplifts? Additionally, on the supply side, how are merchant partners leveraging AI tools to drive demand, improve sell-through, and how are driver partners utilizing AI to optimize their earnings and manage the workload and rest cycles? This is a question for Anthony.
Thanks so much, Alicia and John. Let me take the AI proof points. On engagement, I want to be straight about what we measure. We do not optimize for time spent in the app. For every AI feature, we ask whether people transact more often, more loyal, do they come back, because that is what you can model. Three numbers from this quarter answer it. Users who shop both GrabFood and GrabMart now account for 43% of food GMV, and their baskets are about 10% larger. Groceries GMV is growing 1.8 x faster than food. GrabUnlimited subscribers transact almost 3 x as often as non-subscribers. AI sits behind each of those in search, in recommendations, in offers we show, but we report the outcome rather than the model, and we will keep reporting the same metrics every quarter so you can track them.
On the supply side, the tools are now at scale. The driver AI assistant is with nearly 500,000 drivers. It shows them where demand will be before it peaks, and drivers who actively use it earn over 10% more per online hour. The same information helps them plan their day, their breaks, because they no longer need to stay online just waiting for the next job. I want to be clear on how we see that. These are tools drivers choose to use, and they decide when and how long they want to drive. On the merchant side, merchants on our Merchant AI Assistant, MAI, earn about 20% more on average. Average merchant earnings grew 15% year-on-year, and active advertisers grew 29%.
Look, I'll be honest, adoption still has a long way to go, and that's why we continue leaning in on AI as a core part of our company DNA.
All right. Thank you, Anthony. Continuing on to that thread, additional questions as well on OpenAI and Meta Muse, which are our recent partnerships and collaborations. Questions as well coming from John of Daiwa and Wei of Mizuho. Once again, another question set for Anthony. These are two separate questions on these partnerships as well, and their broader AI benefits. The first component is, what tangible benefits do you expect the OpenAI partnership and your broader AI investments to deliver to Grab? The second component is then on Meta Muse. Grab was announced last week as an early connector partner for Meta Muse in Singapore across GrabMart and Grab Dine Out. Can you discuss the early use cases you are exploring, and where do you see AI agents becoming a meaningful incremental customer acquisition channel for Grab over time? Questions for Anthony.
Thanks, John and Wei. On OpenAI specifically, our OpenAI partnership delivers tangible benefits because we put the tools in our partners' hands, and we teach them how to use them and how to make more money with them. The results are already measurable. As I shared just now, merchants using our AI assistants see about 20% uplift. Drivers who use it earn over 10% more per online hour. To scale, in September, we launched GO Forward with AI under GrabAcademy. Over the next two years, we will equip 30,000 drivers and merchant partners across Southeast Asia with very practical AI skills. I actually met many of them, and I saw how they're using even Imagen to help them become even greater creators, social media creators, to help them boost their income.
We started in Singapore, launched in Thailand in October, and expanding to several other countries, including Indonesia and the Philippines this year end. Partners who use AI sell more, earn more, and stay longer and more loyal to our platform. That gives us stronger supply, as Peter talked about, and better service quality. On Muse, Meta chose us as a launch partner when they brought Meta Muse to Singapore last week, and we deliberately chose GrabMart and Grab Dine Out. Groceries are habitual. People buy much the same basket every week, and that's exactly the kind of task an agent does very well. Dining out is the other everyday use case. Here's how it works. A user asks Meta Muse in chat. Meta Muse would build a basket from a GrabMart store, and the basket shows up in their Grab app, where the user approves the purchase.
Our merchant, our inventory, and the transaction data all stay with Grab. This is our pilot in one market, so it is too early to size the impact on customer acquisition or GMV. We see it first as incremental demand, just a new front door into Grab. We will go partner by partner with platforms that bring real scale and utility, and add services and markets when the time is right. Over time, I do expect agent-originated orders to become a growing channel, and we will share updates as the partnership progresses. What does not change is where the value sits. Whichever front door an order comes through, the merchant network, real-time inventory, dispatch, payment, and customer care all run on Grab. We will also keep improving our own AI experiences so users have good reasons to come to us directly.
Thank you, Anthony. We received a few questions on financial services, so I think we will take them as different components. The first component is on the financial services EBITDA breakdown for the third quarter. This is a question for Peter, coming from several analysts, including Piyush of HSBC and Jacqueline from CGS. For financial services, Peter, can you give us a split of the EBITDA contributions from Superbank, Stash, GrabFin, and the remaining business in the third quarter?
Yeah. We do not split them out, each of those components, because the way we think about it, actually, financial services is one bucket of different components of the ecosystem working together. Hence why we are very, very deliberate of not breaking it out, including the banks itself. What is publicly available, obviously, we will disclose, such as Superbank, which is a public company itself. Just stepping back here, just to remind everyone here, the EBITDA target that we are giving out for 2028 for financial services is $500 million. Financial services today, third quarter, produced $12 million. So we still have some ways to go to get to $500 million. So where is it going to come from? What are the mix that we are going to see?
What we said was $300 million out of the $500 million will come, we believe, from Atome and Superbank, and we said that a few weeks ago. The remaining $200 million come from all the other products, ecosystem of financial services, including the banks, the GXS, GXBank banks, as well as our non-bank lending portfolio, which is coming from what we call GrabFin. We do expect that our banks, over time, given that they are in a scaling mode, to take up a larger share of the gross loan portfolio over time. That would be a great leading indicator in terms of the profitability trajectory that we have, in addition to the Atome also, once the acquisition is closed, given that their loan book also is growing at a very rapid pace at the same time.
In this quarter, actually, we gave out some additional disclosure on financial services regarding NPL, regarding also risk-adjusted returns, revenue-adjusted returns here, to give the community a bit of a sense in terms of how we think about the unit economics of our lending portfolio to help you be able to model actually how we get to the $500 million.
Thanks, Peter. For the second component on the financial services question, this is more in terms of the building blocks to get to the $500 million. Question comes from Alicia, Citi, and Piyush of HSBC. I will turn the question now to Alex to explain this. The question is, as you work towards the 2028 financial services targets of $500 million EBITDA and the above $6 billion loan book target, where do you believe you are today relative to that journey? What are the building blocks required over the next two years in terms of customer acquisition, lending penetration, product expansion, funding costs, and credit performance as required to get you there? Which assumptions do you believe is the most critical in achieving these targets, and under what scenarios could the business meaningfully outperform or fall short of your current ambitions?
Thanks, Doug. We are very proud that it is the first profitable quarter for financial services. I think it was September 2022 when I stood up in front of you and said that we could break even by the end of this year. But here we are in the third quarter, and it is a strong contribution also, $12 million of segment adjusted EBITDA. So up $40 million year-on-year. We are at a very good trajectory in terms of the time at which we are breaking even. We have some, as you know, some significant advantages in terms of how we think about the strategy. Just to summarize what those are for those of you who have not been following our strategy closely on financial services. First of all, the acquisition costs are very low.
We have daily interactions through the super app to 57 million monthly transacting users, and we have annual transacting users far north of 100 million. So it is a very big, significant reach. We have 647 million repayment touchpoints through which we can feed the AI-driven underwriting model. That is a significant advantage too. We can pre-collect many of the repayments from revenues which are flowing through the platform. On the funding side, the reason why we have the banks is that we expect them to be able to give us access to lower cost of funds over time. We have 2.7 billion of digital bank deposits already. Of course, we are building syndicated facilities that we can do in cooperation with Atome over time also, which will give us access to wholesale bank funding.
We are doing a lot more on funding than we were in the past, where we were just relying upon the equity that we hold on the balance sheet. Moving forward, as Peter was just saying, the banks continue to grow their share of the gross loan portfolio. It is a bank group which is structured with a lead regulator in Singapore, so international investors can feel confident of the compliance and the strength of that structure for the banks. Gross loan portfolio growth obviously is outpacing cost structure significantly. We are coming through that break-even point now, so this is where the operating leverage really kicks in, and the returns from the financial services segment will start to accelerate from here. The proposed acquisition of Atome adds consumer lending on a massive scale.
Most of our lending so far has been to drivers and merchants, but now with Atome coming in, these more mature consumer lending models like open-loop BNPL, which it would have taken us many years to develop, are now available to us to distribute immediately, and the network of 30,000 retail brands that they work with already also available to our platform, not just for financial services, but also for advertising and partnering on e-commerce. So, very well- positioned for future growth. We are focused on the partners, as I said, historically, but now this addition of this very large TAM for consumer lending comes into sight. That gives us confidence on line of sight of the $500 million EBITDA target for 2028. So financial services already becoming a major driver of our growth in EBITDA to that total group number of $1.7 billion in 2028.
At that point, the loan book will be above $6 billion, including Atome, obviously subject to completion. But that is still just a medium-term milestone. In my view, as we mentioned at the time that we announced Atome, the consumer TAM here is much, much larger, and so we could be 10 x higher than we are in terms of consumer penetration than where we will end this year, 2026, with our $3 billion loan book. So, lots and lots of runway for us to grow this business and make significant contribution to Grab's future growth and future market cap.
Thanks, Alex. We move on to the next question, which is on Indonesia, a question also from several analysts, including Navin of UBS, Piyush of HSBC, Thomas of Jefferies, and John from Daiwa. This is a question for Alex. In Indonesia, can you share if there is any update on potential regulatory intervention in the delivery segment, and what could be the likely impact of the regulations?
Yeah, as I mentioned in my remarks, the written remarks accompanying the release, we are very concerned about the plight of the drivers faced with the high fuel costs that we are seeing globally. We have been leaning in heavily to make sure that their income keeps pace, and in addition, we have been leaning in with additional driver incentives as well to support them during these difficult times. We are working closely with governments also. In the case of Indonesia, the Indonesian ministries, including KOMDIGI, the Ministry of Transport, the Ministry of Manpower, as the government evaluates the regulatory frameworks for on-demand logistics and food delivery. I think our counterparts in the government truly appreciate that our platform and the other delivery platforms provide earning opportunities for those millions of everyday entrepreneurs in Indonesia. Any regulation that is implemented must not damage those earning opportunities.
We have to proceed carefully and in partnership with government. Our understanding is these guidelines will focus on helping the driver earnings and fee transparency, and we will focus on that with them. We will continue the constructive dialogue. We are not in a position to reveal any new details, but the details are starting to emerge, and as we hear about those, we will let you know as soon as possible. We do not expect any of the business implications of those new guidelines to materially impact our guidance outlook.
Thanks, Alex. Next question is on capital allocation. Also a question from several analysts, including Divya of Morgan Stanley, Gemma of Barclays, Hussaini of Maybank, Sachin of Bank of America, and Jacqueline of CGS. This is a question for Peter. Peter, following the rejection of the foodpanda Taiwan acquisition, are you exploring alternative acquisition opportunities? How would you weigh these against further share buybacks? How much incremental, in addition to that, with regard to the financial services segment, how much incremental corporate capital do you need to scale the loan book?
Yeah. Look, we are disappointed with the outcome of foodpanda Taiwan acquisition. We were anticipating it. At the same time also, we respect the decision made by the TFTC in Taiwan. It also adds another $550 million of additional cash to the balance sheet. From an M&A and capital allocation perspective, let me just frame it this way. When it comes to M&A discipline, we are very anchored on investing first in our core mobility, deliveries, and financial services business. You have seen that track record that we have been doing. We couldn't get there in terms of Taiwan, but hopefully with Atome, which Alex spoke about, is the next one that we feel very, very passionate about. It really opens up a great opportunity in terms of ecosystem. It is also a sizable acquisition that we are doing.
At the same time also, we are working with the regulators just to make sure also that this one has the comfort that we want to get through by the time next year. Now, at the same time also, we have to invest on the organic business, and that is what we have been doing. You saw that in the Q3 results. We talked a little bit about the guidance for Q4 and how we think about the fiscal year also. Alex spoke about GrabMall and GrabMart, how we are investing that also at the same time. We are going to continue to lean in around affordability, which is really important. When you see the MTU growing at 20% year-over-year basis, reaching 57 million MTUs, we added 3 million Southeast Asian into the platform itself, shows that we have to continue to lean in giving access to Southeast Asians to use our products.
It has to be affordable and reliable and accessible at the same time. At the same time also, we have to lean in and invest organically on our partners' earnings. Alex spoke about that, and Anthony spoke about whether it is related to AI or whether it is related to fuel subsidies or whether it is related to the way they are able to earn also in working with our regulators. That is really critical, and we are not going to stop. We are going to continue to lean in to make sure our merchants and our drivers continue to earn more on the platform. We are also continuing to make sure that our products, when it comes to lending, is accessible to our public as well as to our partners at the same time. Atome will unlock that at greater scale for us.
That is going to be a very perfect example where scale can multiply. Alex is so dead on. We are just scratching the surface with consumer credit. The TAM of consumer credit is 10x what Alex has spoke about, and we think it is more than that. So we are very, very eager and keen to actually unlock that box in consumer credit in a prudent way in Southeast Asia. We will continue to invest in that. Now, we are also always looking at long term. Part of my capital allocation strategy in working with Anthony Tan and Alex here is to think long term for the business. We have continued to see where can the business continue to grow and evolve over the next 5 - 10 years. We have been making those bets. You said in autonomous vehicles, for example.
We are increasing cars in Singapore in anticipation of the introduction of a widely more usage of autonomous vehicles. We are adding another 40 cars of autonomous vehicles. We are making those investments. We are making investments in robotics also at the same time. We are making investments in AI also at the same time. So those are more longer term bets that we will use our balance sheet to make sure that those long-term bets are also important, but also prudent at the same time. Now, excess capital will always be there, and we know that. All of you are watching the balance sheet all the time. So we are very focused on shareholders' value. So what do we do? September, October, we accelerated the share buyback, and we repurchased $400 million in those last two months.
I am committed, so is Anthony and Alex, to make sure that we complete the remaining $500 million of our $900 million program over the next 12 months. If we can do it earlier, we will do it earlier. We intend to make sure that that $900 million gets done by September of next year, and we intend to cancel all those repurchased shares also as part of that program. We do generate cash in the business. You see it in the numbers, $533 million. We almost doubled the free cash flow, and we will continue to use that cash to good use. I can assure you to our shareholders that I am very conscious that the balance sheet that we have is a privilege.
We will make sure also that our shareholders return when it comes to the balance sheet itself, and we will make sure that we balance sheet also continues to be very healthy in a net cash position.
Thank you, Peter. The next question is from Venu of Bernstein, and this is a question for Anthony. Question here is, would Grab be willing to partner with global firms such as Uber to minimize long-term risk?
Thanks, Venu. Let me say what I take the long-term risk to be, so I answer the right question. The risks that matter globally over the next 10, 20 years are technology shifts. We talked about AI agents, AI agents changing how people transact, autonomous vehicles changing how people get around. Our answer to both is the same. We partner with the best in the world to bring their technology, we localize it in Southeast Asia on our platform. On Uber specifically, you called out. Uber has been a shareholder since 2018, and Dara and I speak regularly. Today, an Uber user traveling in Southeast Asia can book a Grab ride inside the Uber app, and a Grab user traveling into the U.S. can book Uber inside ours. That works for travelers and brings demand to both platforms. We take the same approach with others.
In AI, as I shared earlier, we work with OpenAI across the region and Meta's Muse in Singapore. In autonomous vehicles, we have invested in WeRide and Momenta to bring robotaxis to the region, and Ai.R, our public autonomous ride service in Punggol, Singapore, is already carrying passengers with commercial service on the Grab app to come. We have partnered for years, and we will continue to do so. What we bring with each of these partnerships is our on-the-ground network, the biggest driver army, merchants, maps, licenses across 900+ cities across Southeast Asia. We are all in on Southeast Asia, and I am convinced that this region will continue to grow strongly in years to come with Grab as its key enabler.
That is why we see today global companies come to us when they want to reach the real Southeast Asia across the region, and it is why new technologies here continues to arrive on Grab.
Thank you, Anthony. The next question is on autonomous vehicle strategy. This question is from Divya from Morgan Stanley and Venu of Bernstein. This is a question for Alex. Can you share your latest thoughts on what does the eventual autonomous vehicle business model look like between Grab, the fleet owner, and AV technology provider? Could AVs ultimately disintermediate Grab, or do you think Grab's demand aggregation, payments, and fleet orchestration mean AV operators will still need Grab as a distribution layer? It is a question for Alex.
Thanks, Divya. You started to, even in the question, you started to outline how we feel about this. First of all, AVs are still in the early stages in Singapore. We have already clocked over 100,000 km in local trials, and we have plans to expand our fleet here in Singapore by fivefold. Currently, we are growing it to 50 vehicles, and we are preparing to launch the commercial autonomous ride hailing on the Grab app in the coming months. The steady state model divides into basically three specialized layers. There is the AV OEM, the technology provider, and they develop and license the autonomous driving stack with the remote guidance software that is necessary to make that run. Then there is a fleet owner or financier. They fund the capital expenditure.
They take the capital, they take the depreciation through their P&L and balance sheet, and then they manage the depot charging, the cleaning, the maintenance. There is the orchestrator. That is primarily Grab, platform orchestrator, demand aggregator. In some cases, Grab can also get involved in depot management where necessary to get things running. Our track record in Southeast Asia has been to develop relationships with those fleet owners and fleet operators. We can continue that model. We ourselves remain primarily asset light under that model. But the fleet operators who currently are using conventional vehicles, so not AV, but the standard driver-driven vehicles, we will work with them to transition into this new world. In some cases, with our banks, we can help them with financing as well. We can add capacity as demand grows.
Although at the start, where AVs currently are very young technology, and we are developing the capabilities to understand how they interact with the ride-hail platform, what kind of depot management and fleet management capabilities are needed, as we scale to massive asset deployment, hundreds of billions or maybe trillions of dollars of AV asset deployment in the long run, then we will combine our ride-hailing technology with this new hands-on AV experience, and then we will be able to optimize the integration with the OEM to get the optimum integration of demand and fleet management, working with regulators, those technology partners, and the communities on the ground, as we always have. Let us focus specifically on Southeast Asia, because AVs cannot handle Southeast Asia transportation alone at the moment, with a lot of weather, huge extreme tropical monsoons, et cetera.
There are lots of operational edge cases where the technology is still developing. Probably more significantly, even once the technology can overcome those, the cost displacement curve for human drivers in Southeast Asia is going to be many, many years. It is not like North America, where the driver cost per hour is much higher than it is here. In most of our markets, it is going to be a very long transition, probably a decade, something like that, where there are drivers and AVs working side by side. That is obviously where the platform orchestration is critical. So we will have a hybrid dispatch orchestration, load balancing between AVs, human drivers. Some passengers will need assistance, so they will have preferences for human drivers. Other passengers might have preferences for AVs, but that is all going to be orchestrated through the app.
Those passengers, of course, are not just using mobility. In the Grab app, the super app, they are using all our other services as well. You can see from our numbers that the intensity of cross-sell between food, mobility, financial services, and now GrabMart and GrabMall as well, is very intense. The customer relationship is very much intact with Grab. We are confident that our goal of scaling AVs with fleet partners across Southeast Asia in the coming years in this asset-light model, in complement with the human driver fleet, is something which will just enhance the position of our platform as the orchestration layer across all of those partnerships.
Thanks, Alex. We have time for one last question. The final question is on GrabUnlimited, and this is a question from Wei of Mizuho. A question for Alex. Subscribers are now over 20% of deliveries MTUs, but they generate 35% of deliveries GMV, and they transact nearly 3 x as frequently as non-subscribers. At this scale, is the larger economic benefit coming from higher order frequency, lower acquisition and retention costs, or is it greater cross-sell into mobility and financial services?
Yeah, thanks, Wei. You are right to highlight GrabUnlimited. It is a critical relationship-building tool for us, which drives primarily frequency and retention. GrabUnlimited users transact over 3 x more than non-users and have significantly higher retention rates. That is the key benefit from the program. They also allow us to reduce promotions, so consumer incentives outside of GrabUnlimited can come down from where they were before GrabUnlimited had such high penetration, and that also helps to fund the program. The benefits today concentrate primarily in deliveries, but we have begun to extend it to mobility benefits in those markets, in particular where we have got very healthy supply situations. That is working very well and has improved our cross-sell and retention on the mobility side as well. Merchants are very interested in this universe of GU users.
They know that those are the ones with the high spend and the high frequency, so they are very important customers for our merchants to have access. A lot of spend on the platform is from merchant-funded promotions, and you can see that that has gone up 48% year-on-year. A meaningful portion of that increase is driven by merchants wanting to reach the GrabUnlimited user base with their exclusive offers. A great time to join GrabUnlimited with all those exclusive offers, but also a very attractive magnet for merchants to drive their success as well. You should expect us to continue to expand GrabUnlimited for deliveries, continue to drive penetration into the other verticals, and you can also expect us to use GrabUnlimited to extend benefits beyond the ecosystem.
We are leveraging, for example, there is a partnership with OpenAI to offer free ChatGPT Plus and Trip.com booking discounts. That is going to be an avenue that you see more and more of also, so we can bring new benefits for our loyal GrabUnlimited customers.
Thanks very much, everyone. That brings us to the end of our Q&A session. I will now turn the time over to Peter for closing remarks.
Thanks, Doug. Look, before we wrap up, I just want to take a moment to acknowledge what this year, the last nine months, has delivered. Adjusted EBITDA reached new record levels in every quarter. Our Financial Services business crossed into profitability, and we have nearly doubled the trailing 12- month adjusted free cash flow. That reflects the durability of the Grab platform itself. In addition also, we back that with capital. So we have repurchased $400 million of shares in September and October, and we reflected it in our guidance, raising both revenue and also adjusted EBITDA for the full year for the second time this year. As I always remind the Grab team, this is not a one-quarter story. It is the trajectory where it counts. These milestones would not be possible without the trust of everyone who relies on Grab. Our drivers and our merchant partners, thank you.
It is what has been a rough quarter for a lot of you, given the weather as well as the fuel prices, so thank you. For our merchant partners also, for the restaurants, thank you for supporting us. And to the millions of consumers who use the Grab services every day, thank you for trusting Grab. Thank you for opening up the app, ordering a ride, [booking], ordering a restaurant for delivery, or a [GrabMart], or a [GrabMall]. Thank you for doing that. And also to all our Grabbers who worked tirelessly in the third quarter and all through this year, thank you very much. Anthony, Alex, and I also want to thank our shareholders also for your continued support. As usual, the IR team will be on the road over the next few weeks, so do reach out.
We will be in a few countries, including Singapore also, so feel free to reach out. We would love to have a dialogue with you or even sit down and have a coffee with you. Thanks, everyone, for joining us today, and see you next quarter.