Pony AI Inc.

Ladies and gentlemen, thank you for standing by, and welcome to Pony AI Inc.' 's fourth quarter and full year 2024 earnings conference call. At this time, all participants are in a listen-only mode. After the management's prepared remarks, there will be a question and answer session. As a reminder, today's conference call is being recorded, and a webcast replay will be available on the company's investor relations website at ir.pony.ai. I will now turn the call over to your host, George Hsiao, Head of Capital Markets and Investor Relations at Pony AI. Please go ahead, George.

Thank you. This is George speaking. Hello, everyone. We appreciate you joining us today for Pony AI's fourth quarter and full year 2024 earnings call. Earlier today, we issued a press release with our financial and operating results, which is available on our IR website. Joining with me today on the call are Dr. James Peng, chairman of the board, co-founder and chief executive officer, Dr. Tiancheng Luo, director, co-founder and chief technology officer, and Dr. Liu Wan, founding member and chief financial officer. They will provide prepared remarks followed by a Q&A session. Please note that today's discussion will contain forward-looking statements made under the safe harbor provisions of the U.S. Private Securities Litigation Reform Act of 1995. Forward-looking statements are subject to risks and uncertainties that may cause actual results to differ materially from our current expectations. Further information regarding these and other risks and uncertainties is included in the relevant public filings of the company as filed with the U.S. Securities and Exchange Commission. The company does not undertake any obligation to update any forward-looking statements, except as required under applicable law. Please also note that Pony AI's earnings press release and its conference call include discussions of both unaudited GAAP information and unaudited non-GAAP financial results. For reconciliation of these non-GAAP measures to the most directly comparable GAAP measures, please refer to Pony AI's disclosure document available on our IR website. I will now turn the call over to our chairman, co-founder, and CEO, Dr. James Peng. Please go ahead.

Thanks, George. This is James Peng, founder and CEO. We consider this is an exciting time for Pony AI as we report our first earnings results as a public company. Our NASDAQ listing marks a significant milestone and is timed perfectly with the imminent mass commercialization of our robo taxi services. With ample financial resources now available, we are well positioned to lead and capitalize on the upcoming large-scale rollout of robotaxis, making this year an inflection point for the widespread adoption of autonomous transportation solutions. We are taking robotaxis first, China first, and the Tier 1 cities first approach. This is our current focus. is on scaling robo taxi operations in China, which not only generates sizable recurring revenue, but also offers a solid foundation for further expansion into various global markets. China's online ride-hailing market is exceptional. The country's tier one cities, mainly Beijing, Guangzhou, Shanghai, and Shenzhen offer a unique combination of demand, consumer readiness, and regulatory clarity, making them ideal for large-scale robo-taxi deployment. We estimate each city can easily support a fleet of over tens of thousands of robo-taxis. With technology that meets regulatory standards and backed by the fully driverless fare charging licenses, we have already secured. We are all ready for quick scale up. Launching large fleet in tier one cities will enable us to validate our business model, optimize our operations, and establish these markets as a benchmark and a scalable framework for future growth, either into other Chinese cities or extends to international markets. Next, I will explain why we anticipate our global taxi service will soon achieve mass commercialization. First and foremost, we have achieved technological readiness for mass commercialization. Our operational records proved that our Robotaxi has achieved level 4 driverless operation 24-7 in all weather conditions, making it commercial ready. Our technology is empowered by Virtual Driver and the World Model. The Virtual Driver is a comprehensive full-stack system proprietary software and hardware. This enables us to collaborate effectively with automakers and transportation network companies, we call them TNCs, to create a scalable robotaxi business model. Additionally, our generated Pony World model treats our virtual driver to be much safer and better than an expert human driver through advanced reinforcement learning. Our pony world simulates a wide range of scenarios, including extreme cases and the long tail events. By employing a training method called learn by practicing, our virtual driver does not just know what to do, it actually understands the reasons behind its actions. This is quite different from imitation learning that is widely used for the typical L2 systems. Because the L2 systems imitate the driving patterns of human drivers, they can only reach human level safety. In contrast, Pony World has improved our virtual driver's safety by 16 times, while at the same time significantly improving its comfort and the driving efficiency. Our safety record enabled our commercial insurance costs per robo-taxi to be reduced to almost half that of the traditional taxis. We have established strong relationships with local governments and secured the required policy approvals for large scale commercialization. We have obtained all the most advanced licenses in China's tier one cities. For instance, in recent weeks, Pony AI launched paid robot taxi service that connects key transportation hubs such as Beijing South Railway Station, Beijing Daxing Airport, and Yizhuang District, with plans to gradually expand to Beijing City Center. Moreover, in February this year, we launched paid robotaxi services in Guangzhou City Center, Guangzhou Baiyun International Airport, and Guangzhou South Railway Station. We are the first and owning company approved to provide robo taxi services on these high demand routes. Moving forward, we'll gradually expand our operations in these cities, paving the way for future growth. Third, we have built extensive mass production partnerships to support large scale commercialization. For example, in the first half of 2024, we established a joint venture with Toyota. As part of the deal, we will roll out mass production of the robotaxis based on BZ4X, as well as build the value chain of autonomous driving operations, including maintenance, charging, and other aspects. In addition, in the second half of 2024, We respectively reached mass production partnerships with BAIC New Energy, that is Beijing Auto, and the GAC Group, that is Guangzhou Auto. Based on the BAC ArcFox Alpha T5 models and the GAC Ion models, we carried out cooperation in the mass production of auto-grade autonomous driving kits, vehicle model production redundant safety design of the chassis and some other areas. These partnerships have been reinforced through strategic equity investments from all these three OEMs. All three upcoming Robotaxi vehicles are based on our seventh generation autonomous driving systems. This latest generation has achieved a major breakthrough in cost efficiency, reducing unit bomb cost by over 70% compared to the previous generation, with further cost reduction anticipated as we scale up. Fourth, we have fortified our operational capabilities to support the ramp up of fleet size and accommodate fast-growing user demands. We have developed our own ride-hailing platform, which is called Pony Pilot, and forged strategic partnership with leading DSCs, such as On-Time Mobility and Alipay, to offer driverless robotaxi services. In the fourth quarter of last year, we also established a partnership with Alibaba's online mapping and ride-hailing platform, AMAP, and integrated our Robotaxi service into its mobile app and meeting programs, making our services more accessible to the public. In 2024, the average daily orders per vehicle reached 15, and in Q1 2025, we continue to see the growth of daily orders per vehicle. with significant progress have been made in all the four pillars of autonomous driving. That is technology, regulations, mass production, and large-scale operation. We do see that a critical inflection point for mass commercialization is right in front of us. Now let's look at our robot truck business. which we have also seen significant growth in 2024. We deepened our joint venture with Final Chance, transforming it into a comprehensive autonomous driving transportation as a service platform. Together, we'll continue building a smart, efficient, safe, and environmental-friendly logistics road transport network while further expanding our robotruck fleet. A major milestone that highlights our leadership in robotruck business is our approval as the first company in China to conduct robotruck driver-out platoning on cross-provincial highways linking Beijing, Tianjin, and Hebei Province. With only the leading truck requiring a safety operator and the following trucks to be fully driverless. Testing has already begun on the Beijing-Tianjin Tanggu Expressway, making a significant step towards full autonomy for all trucks in the Platon, which will further reduce logistics costs and accelerate commercialization. In summary, Our transition to a public company marks the beginning of an exciting new chapter. We stand at a defining moment as we move toward the large-scale commercialization of autonomous mobility and continue to gain momentum, building on a solid foundation of technological advancements, regulatory support, and industry partnerships. Looking ahead, Our priority for this year is clear. Accelerating the mass production and the deployment of our seventh generation Robotech fleet. Further reducing the unit bomb cost and expanding operation areas and density in China's tier one cities. With that, I'll now pass it over to our CTO, Dr. Tiancheng Luo to review our technological progress. Tiancheng, please go ahead.

Thanks, James. Hello, everyone. This is Tiancheng. So I'm delighted to have this opportunity to share with you the latest progress of our technologies. Pony's technological development is centered around enabling the mass commercialization of robotics. To achieve successful lobotactic controlization, autonomous driving technology must meet three key criteria. First, it must attain a sufficiently high standard of safety. Over-experience shows that a magnitude safer than a typical human driver is attainable and should be needed. Secondly, cost control is essential. Cost should be managed across various aspects. including sensors, computing hardware, daily operation, and insurance. Low cost ensures that Robotech's service remains economically sustainable. Finally, Robotech's service should cover a large enough geographical area to enable large-scale operations. According to our operational and safety records, Tony's technology has matured to a level that can support mass commercialization. focusing on safety, cost-effectiveness, and intensive service coverage. Through years of effort, we have been commercially operating fully-gravity robot taxis for over two years. Even this time, safety has already surpassed typical human driver by an order of magnitude. As we progress, costs are expected to decrease by 70% in next generation, which will be mass-produced in the second half of this year. Moreover, our service coverage has received regulatory approval licenses in all tier 1 cities in China, which are capable of operating tens of thousands of robo-taxis. Moving forward, our technical goal will remain focused on enhancing cost efficiency and operational capability without compromising safety. In the competitive landscape of robo-taxis services, only companies that can run driverless commercial operations with a significant fleet, hold a position at the forefront. Years of innovation and diligence have given us a strong competitive edge. It took us four years to progress from initial jobless demonstration to fully launching commercial robot tech services in China and Taiwan cities. You may wonder why it took companies like Waymo and Pony almost five years to get there, from demo to commercial operations. The reason is that we had to move from simply matching human driving capability to significantly exceeding them. This means we had to rebuild over core algorithms as all the ones were designed in a way subject to human limitations. Now let me further explain why the technology evolution that allows us to bridge the gap and launch over fully driverless services. The key is moving from imitation learning to reinforcement learning. A trend that is the key driver brought us a seat at the forefront. With imitation learning, which is still widely used by most of the L2 systems, the AI drivers learn by copying human behavior using data from the real world driving. By mimicking human driving patterns, imitation learning cannot understand the reasoning behind the driving behavior. As a result, the solution is not general enough to handle ever-changing traffic scenarios Reinforcement learning, on the other hand, uses a generative virtual environment called a word model, or pointy word as our team likes to call it, where our virtual driver teaches itself through billing of even trillions of generative acts of trial. This allows our virtual driver to understand why by analyzing the outcome of every action, equipping them to make smarter decisions in complicated scenarios. Through repeated reinforcement learning, our virtual driver gradually learned to adapt to new situations, unexpected challenges, and corner cases, preparing them to operate in the real world. Over time, our virtual driver trained on the pony world developed the advanced skills needed for complex tasks, such as smoothly navigating busy streets, handling unpredictable traffic scenarios, of safely operating for tens of thousands of hours without any incident. There are three key components making our Pony World approach possible. The ability to generate realistic scenarios and sensor data, a high-fidelity simulation system, and a comprehensive set of evaluation metrics. Together, these elements allow our Pony World to effectively coach our virtual driver to handle real-world challenges. I would like to highlight our high-fidelity simulation engine here, which leverages the latest technology to create an environment that precisely replicates real-world conditions in both subtle details and dynamic responses. Unlike traditional systems that rely on human driving data, our simulation engine generates its own driving scenarios and challenging situations for autonomous vehicles to understand, adapt to, and make decisions in. The traffic participants in our simulation engines are designed to behave like real humans, interacting with autonomous vehicles in a natural and human-like way. This makes our Pony World a powerful tool for coaching over virtual drivers. Finally, let me share the latest progress we've made in advancing our technology for mass production and commercialization. Large-scale commercialization requires handling lower probability extreme cases with hardware that has lower performance. To address this challenge, we continue to innovate our pony world. Here's how it works. We have trained an Oracle AI driver in our pony world, a virtual environment that can be rewound. This Oracle learns to predict future outcomes and then acts as a coach to train other AI drivers, helping them anticipate and respond to future events. Using similar methods, we have been able to maintain safety standards for mass-produced and auto-grid LIDAR domain controllers and larger robo-taxi fleets. Tony Ward has improved our virtual driver safety record by 16 times, while significantly improving its comfort and driving efficiency. This advancement has reduced the commercial insurance cost per robo-taxi to almost half of that of traditional tactics, as this is a clear objective measure by the insurers of safety of our technology. Before I conclude, I would like to highlight the creation of Pony World took years to dedicate its research and development, driven by a team of exceptional talented engineers who evolved and thrived together with us over time. This journey was fueled by the belief that our pony world offers greater potential and is critical for achieving driverless robotics commercialization. Those years we spent were the toughest for our company and for me personally. I deeply grateful for the trust and the support of our investors and colleagues along the way. This concludes my prepared remarks. I will now pass the call over to our CFO, Dr. Leo Wang, for a closer look at our financial results. Leo, please go ahead.

Thank you, Tianchen, and hello, everyone. I'm pleased to present Pony AI's financial results on our inaugural earnings call. Looking back on 2024, we kicked off our seventh-generation automotor driving system development, with three OEM partners, which is critical to execute our Robotaxi First, China First, and Tier One City First strategy. We also deepened the partnership with industry leaders, creating a robust ecosystem that accelerates the adoption of these technologies. During our IPO late last year, we raised over 400 million US dollars which provided us with ample firepower to drive our strategy. Looking forward, we'll concentrate and accelerate our seventh generation autonomous driving system development and deployment in China's tier one cities, hence to solidify Pony AI's position for sustainable growth. Moving to our financial performance, please note As we navigate the early stages of commercialization, we are experiencing volatility in our quarterly revenue and margins, which is expected to continue in the near term. But we are focused on executing our go-to-market strategy and achieving key milestones laid out by James and his remarks. which we expect to reduce variability in our financial performance in the future. Now let's take a closer look at our financial results for 2024. For additional quarterly results, please refer to our earning release, which is posted online. Our full-year revenue totaled $75 million, an increase of 4.3% year-over-year. RoboTaxi services revenue was $7.3 million, down 5.3% year-over-year. The decrease was primarily driven by reduced service fee from providing autonomous vehicle engineering solutions based on our project progression schedule. Our RoboTaxi services revenue also included passenger fares, which saw a significant year-over-year increase driven by the expansion of our public-facing fare-charging road taxi operations in Tier 1 cities. We expect this part of growth will continue and even accelerate as we deploy the seventh generation of town-driving vehicles. Robotruck services grew strongly, delivering $40.4 million in revenue, up 61.3% year-over-year. This robust growth was driven by the expansion of our fleet into new regions, where new demands can be fulfilled by our robotruck fleet. Licensing and application revenue was $27.4 million, down 30.1% year-over-year, influenced by recognition schedule of project-based revenue. Total cost of revenue was 63.6%. million US dollars, up 15.6% year over year, in line with revenue trend and revenue mix. We achieved gross profit of 11.4 million US dollars, resulting a gross margin of 15.2%, a decrease from 23% in 2023. The year over year decrease was mainly due to services with relatively low gross margin contributed increasingly to our revenues. Moving forward, we expect gross margins to improve as we further scale and optimize operation over time. Total operating expenses were $296.9 million, an increase of 85.4% year-over-year. Excluding share-based compensation, Non-GAAP operating expenses were $169.9 million, up 8.7% year-over-year. The increase was mostly driven by accelerated R&D investment to support the launch of our seventh generation robotaxis vehicles in collaboration with our OEM partners. Loss from operations was $285.5 million, compared to 143.2 million US dollars in 2023. Non-GAAP loss from operations was 158.5 million US dollars compared to 139.5 million US dollars in 2023. NAS loss was 275 million US dollars compared to 125.3 million US dollars in 2023. Non-GAAP net loss was $153.6 million, compared to $118.5 million in 2023. Turning to our balance sheet, our combined cash and cash equivalents, restricted cash, short-term investments, and long-term debt instruments for wealth management was $825.1 million at the end of 2024. And lastly, for our business outlook. As mentioned earlier, we expect continued fluctuation in our quarterly revenue, as well as margin, since we are at the nascent stage of commercialization. While we're not given formal guidance at this time, we are confident in our ability to scale up commercialization, drive sustainable growth, and deliver value to our shareholders. I will now turn the call over to the operator to begin our Q&A session. Thank you.

Thank you. We will now begin the question and answer session. To ask a question, you may press star, then 1 on your telephone keypad. If you are using a speakerphone, please pick up your handset before pressing the keys. To withdraw your question, please press star, then 2. For the benefit of all participants on today's call, please limit yourself to two questions. And if you have follow-up questions, you may re-enter the question queue. If you ask your question in Chinese, please repeat them in English. The first question today comes from Verena Jang with Goldman Sachs. Please go ahead.

Thank you, management team. I have two questions. My first question is about the business strategy. So what's the strategic rationale behind your robot taxi first, China first, and also the tier one cities first approach? If you could share more color behind, this will be appreciated. Thank you.

I'm James Peng. I'll take the first question regarding our 3.1 strategy. Actually from day one that Tony was founded, autonomous mobility everywhere has always been our company model. This model actually reflects our vision to bring autonomous transportation to all global markets and across all types of vehicles. We certainly have the ambition for other markets down the road. The fundamental reason behind our China First, Robotaxi First, and Tier 1 Cities First strategy, lighting our confidence in an imminent opportunity for mass commercialization. China has the largest ride-hailing market with around 40% of the global market measured by the number of orders. This is roughly twice the size of the US market. Within China itself, Tier 1 cities represent the largest share, backed by supportive regulatory environment and growing users' demand. In 2024 and 2025, we expanded our operations of paid robo-taxi to more railway stations, international airports, and the city centers in Beijing, Guangzhou, and Shenzhen. We also observed that China has established the regulatory framework for robotaxis in a swifter and more transparent manner compared to many other regions. As a result, we believe the Tier 1 cities in China are ripe and ideal for mass deployment of robotaxis. Not only is robotaxis representing the largest market It is also representing the most difficult technical and deployment challenges. The safety requirements in handling the bad weather conditions such as rain and snow and other unpredictable corner cases are very challenging. We have proven our capability to handle such challenges by successfully operating fully driverless global taxis in the last two years. It is from a commercialization perspective that we are currently more focused on robotaxi in China's tier one cities. But certainly, our know-how can enable us to transfer to other transportation modes and also the global market in the future. So thank you. Now back to the second question.

Thank you, James. My second question is about a business model. Could you differentiate your business model against the OEM ride-sharing company and also the taxi company, and any collaboration with these companies? Thank you.

For this question, I think our CFO, Leo, is the right one to answer it.

Thank you, James. I'll take this question. So for our Robotaxi fare charging service actually is focusing on providing a virtual driver who takes charge of the driving in the transportation service. And if you look at the traditional transportation service, that's actually provided by a human driver to take charge of the driving. And we charge our passenger based on the distance driven by our virtual driver. During the ride, actually, we provide a more private and safer experience to the passengers. So if you look at this business model, you can regard that as an upgrade to the current ride-hailing business model, not a disruption. From a ride-hailing platform company perspective, its business still will be matching passenger demand with driver resources, in which you can consider our virtual driver to be part of the driver pool. Automakers, or OEMs on the other hand, they get revenues from selling purposely-built vehicles that are co-developed with Pony. And these vehicles will be sold to robo-taxi operators, for example, Pony itself. In a nutshell, actually, each party in the value chain in ride-hailing business will still play its role in the transportation mobility service sector. We consider this will be a win-win concept. And because this concept is not only supported by us, but also supported by our partner, you can see that we have secured mass production plans with OEM partners, such as Toyota, Beijing Auto, and Guangzhou Auto. We have also integrated into different traffic map companies, such as AMAP, Alipay, OnTime, and et cetera. my answer to your question. Now I will turn back to the operator.

The next question comes from Ming Sun Lee with Bank of America. Please go ahead.

Thank you for giving me the opportunity to ask a question. So my first question, do you foresee any challenges before mass commercialization? Maybe we can elaborate more in terms of the user acceptance technology, maturity, and regulation. Thank you.

I'm James Peng. I'll take this one first. Thanks, Vincent. As I described in my opening remark, I'm very confident that the four key pillars for the mass production of Robotexy, namely the technology, regulation, mass production, and the large-scale deployment, are actually all in place for Pony. I particularly want to emphasize that our technology has advanced the safety of our Robotaxis to a level that actually allows us for the large-scale commercialization of Robotaxis. We do not foresee any insurmountable challenges that prevent us from achieving mass commercialization. Currently, we work hand-in-hand with OEMs and the supply chains to launch a new generation of cost-effective robot taxis, successfully reducing our unit cost by 70%. Along with continued improvements in operational efficiency, we are now on the right track to achieve break-even at the individual vehicle level. In other words, we will have a positive contribution margin from the seventh generation robo-taxis. In general, we have seen supportive regulatory environment from both the central and the local governments. We take pride in being among the first companies in China to secure licenses for operating fully driverless robo-taxis across all four tier one cities. Furthermore, we are the only autonomous driving technology company that has obtained all the necessary regulatory permits required to offer commercial public-facing robot taxi services in Tier 1 cities. Moving forward, our main priority will be expanding our fleet size, operational areas, and the vehicle density to scale up revenue and enhance our profitability. So that's the answer to your first question.

Thank you, James. So my second question, what are the key technological milestones that need to be achieved to enable your mass production of robot taxi service in 2025? Thank you.

Thank you, Mingxuan. I think this one is related to technology. So I'll hand over to Kianchen to answer it.

Yeah, sure. This is Tiancheng. So, as I described in my remarks, SoPony's technological development is centered around enabling mass commercialization of robotaxi. To achieve successful robotaxi commercialization, autonomous driving technology must meet the three key criterias. They are safety, cost-effectiveness, and intensive service coverage. So, through years of effort, we have been commercially operating fully driverless robot taxis for over two years. During this time, safety has already surpassed typical human driver by an odd of magnitude. And cost-wise, as we progress, costs are expected to decrease by 70% in the next generation, which will be mass-produced in the second half of this year. Moreover, our service coverage has received regulatory approval and licenses in all Tier 1 cities in China. which are capable of operating tens of thousands of robot taxis. So according to our operational and safety record, we believe Pony's technology has matured to a level that can support mass commercialization. And moving forward, our technical goal will remain focused on enhancing cost efficiency and operational capability without compromised safety. Yes, thank you, and back to the operator.

The next question comes from Ben Wong with Deutsche Bank. Please go ahead.

Thank you. Actually, just one question about technology. How do you achieve a very high safety level compared to human driver? And why do you believe the level for autonomous driving technology depends on more you generate high quality data rather than the massive data you get from the street? Thank you.

Tiancheng, this one is still yours.

Sure. Yeah, so this is Tiancheng. So yeah, good point. Let me reemphasize that the L4 AI driver is trained using reinforcement learning in a virtual world where data is generated. So as a result, reinforcement learning does not require a huge amount of real-world data. Let me further elaborate on why using real-world driving data to mimic human driving behavior cannot meet and full safety requirements. The fundamental reason lies in the double standard applied to human drivers versus AI drivers. Sociality holds AI to a much higher standard than human drivers. People are far less tolerant to AI mistakes, where AI is perceived as a machine expected to eliminate human shortcomings. This creates a paradox. L4 systems must meet safety expectations far beyond that human driver can achieve. Imitation learning, by its nature, is limited by the feeling of human performance, and the cells cannot satisfy with safety requirements. So although the amount of data used for imitation learning driving is extremely large, it still cannot ensure that the driving capability can surpass that of humans. Another important factor is that Leveraging real-world data cannot understand the reasoning behind driving behavior because it only mimics the driving path of human drivers. There's a common saying that describes this phenomena. One knows what state it is but doesn't know why it is so. Merely mimicking the action of human drivers does not guarantee understanding of the reasoning behind this action. So in summary, compared to most L2 systems, L4 systems use a different data solution where generative data is the key, not the real-world data. Thank you. Thank you.

Thank you.

The next question comes from Purdy Ho from Hoi Tai Securities. Please go ahead.

Hello, management. This is Purdy Ho from Hoi Tai. Hello. And my questions are from Leo, I guess, because most of them are on cost and revenues. So would you mind giving some colors on why your 2024 costs and expenses were higher year over year, and also any guidance on costs going forward that you can provide?

So I'll take this question. So as I mentioned in my earlier remarks, actually excluding share-based compensation, our non-GAAP R&D expenses were $137.8 million. It represents an increase of 14% compared to $120.9 million in 2023. That's mostly because since the second half of 2024, we have been working on three vehicle models of our seventh-generation automobiling system. This incurred the corresponding R&D expenses growth. We consider this ongoing development is very critical to implement our RoboTaxi First, China First, and T1 City First strategy. And much of this development work will be accomplished in this year. So on the other hand, during our IPO late last year, we raised over 400 million US dollars. Now we have a strong balance sheet with a total of 825.1 million US dollars. combined with cash-cash equivalents, restricted cash, short-term investment, and long-term debt instruments for wealth management as of December 31, 2024. This provides amplified power to execute our strategy, but also as a startup, we still need to carefully manage our resource allocation and investment to seek for the best efficiency and returns. So we will continue and even accelerate our seventh generation system development as our top priority and deploy these vehicles in tier one cities from hundreds to thousands. Therefore, we expect the corresponding expenditure will continue to this year. So get back to you.

OK, sure. Yeah, I got a follow up. So we also noticed that revenues the entire year 2024 was up, but for the quarter, the fourth quarter, revenue was down. So any comments on that and any guidance going forward? Thank you.

I'll continue to take this question. So if you look at our current revenue, it consists of recurring revenue, such as we provide a robo-taxi fare charging service to the public. We also provide robo-track logistics service to our business partners. And we also have so-called project-based revenue. For example, launching a proof of concept robo-taxi fleet with our partner for a certain amount of time in certain markets. Given we have a portion of revenue that is tied to milestone-based projects, so that revenue will be recognized upon delivery of contractual obligations. Revenue recognition naturally would fluctuate across different quarters. This is very common for this type of revenues. But look, we are focusing on our seventh generation autonomous vehicle development and the deployment. And these new and more cost effective vehicle will be put into global taxi fare charging operations later this year, starting from hundreds to thousands. So as such, I think it will increase the recurring revenue portion and gradually to change our revenue mix. Hence, we think we can mitigate the fluctuation in the revenue stream in the future. And we also expect our overall revenue will gradually to grow in the near term, following the revenue trajectory in the recent years. So this is my answer to your question. Back to the operator.

The next question comes from Xiaoyi Li with Jefferies. Please go ahead.

Thanks for taking my question. I have two questions. My first question is from the technology perspective. Given a recent emergence of disruptive technologies like DeepSeq, how do you see these play out in the development for autonomy from the industry level? Will they have a positive or negative impact on your technology urban map? And also, will they impact Pony's timeline for the mass deployment of Robotaxi? This is my first question.

Tiancheng?

Yeah, thank you. This is Tiancheng. I can take this one. First, I would say let's broaden the topic a little bit. In the past few years, many disruptive technologies have emerged, including end-to-end architecture, transformer, and also other technologies using the DeepSeq. They all are giving companies like Pony a greater advantage. For example, the integration of the end-to-end technology has significantly enhanced Pony's service coverage. More importantly, the successful commercialization of RoboTaxi involves multiple factors, The destructive technology can only impact one of them. For instance, the factors include, number one, driving capability, such as safety, comfort, and efficiency. Number two, cost, such as sensors, computing operations. And lastly, partnerships, such as OEM suppliers, TNCs. The key to success, so a successful commercialization of Global Taxi is ensuring that all these factors meet a certain standard. Destructive technology can only affect one aspect, and any singular breakthrough only provides marginal help for the entire autonomous driving system. Yeah, thank you, and back to you.

Thank you. That's very helpful. And my second question is regarding your cooperation with OEMs. Could you please share more details about the current progress. How does such partnership help you to achieve your mass production goals?

This is regarding partnership, so I'll take it. I'm James. So our deep collaboration with OEM is one of the keys to actually ensuring our Robotexy commercialization at scale. We work closely with OEMs to co-develop and produce autonomous vehicles across various vehicle platforms. Most importantly, as I mentioned earlier, scale will be instrumental to enabling us to achieve positive unit economics at a quicker pace. Our collaboration with OEMs are set to significantly reduce unit costs, slashing the bomb by 70% compared to our sixth generation robot taxi. In 2024, we have reached agreements with three OEMs, Toyota, Beijing Auto, and Guangzhou Auto, to produce three new vehicle models. Our collaboration with BAIC and the GAC will also endow us with more robust government support our key markets. So in summary, I think our partnership with OEMs actually also goes beyond manufacturing. For example, the joint venture we established with Toyota last year is actually have a more comprehensive partnership. It will provide capitals for vehicles, operate as a fleet company to burden the capex, and also utilize the existing Toyota dealer network for the vehicle maintenance. So that's the answer to the question. Back to the operator.

There are no further questions. Now I'd like to turn the call back over to management for closing remarks.

Hi, this is George Chao again. Thank you everyone once again for joining us today. If you have any further questions, please feel free to contact our IR team. We look forward to speaking with you in the next quarter.

This concludes the conference call. You may now disconnect your lines. Thank you and have a great day.