Agora, Inc.

Thank you for standing by. Welcome to Agora Inc's first quarter 2024 Financial Results Conference call. At this time, all participants are in a listen-only mode. After the speaker's presentation, there will be a question and answer session. To ask a question during the session, you need to press star 11 on your telephone. You'll then hear an automated message advising your hand is raised. Please be advised that today's conference is being recorded. The company's earnings results press release, earnings presentations, SEC filing and a replay of today's call can be found on its IR website at investor.agora.io. Joining me today are Tony Chow, founder, chairman and CEO, Ching-Po Wang, the company's CFO. Reconciliations between the company's GAAP and non-GAAP results can be found in its earnings press release. During this call, the company will make forward-looking statements about its future financial performance and other future events and trends. These statements are only predictions that are based on what the company believes today and actual results may differ materially. These forward-looking statements are subject to risk, uncertainties, assumptions and other factors that could affect the company's financial results and the performance. of its business and which the company discussed in detail in its filing with the SEC, including today's earnings press release and the risk factors and other information contained in the final prospectus relating to its initial public offering. Agora Inc. remains no obligation to update any forward-looking statements the company may make on today's call. With that, let me turn it over to Tony. Hi, Tony.

Thanks, Peter. and welcome everyone to our earnings call. I'll first reveal our operating result in the past quarter. Agora revenues were $15.8 million in the first quarter, up 3% quarter over quarter. This is a great result considering the macroeconomic challenge under the high interest rate environment and is mainly driven by usage growth from emerging use cases such as live shopping, As of the end of this quarter, Agora had over 1,700 African customers, up 16% compared to one year ago. Xiong Wang's revenues were 122.6 million RMB in the first quarter, down 16% year-over-year, mainly due to challenging macroeconomic and regulatory environment. as well as the disposal of the customer engagement cloud business in the first quarter of 2023. As of the end of this quarter, Shonwon had over 3,800 of active customers, down 2% compared to one year ago. Now moving on to our business, product and technology updates for this quarter. Despite a challenging operating environment, we continued to focus on enhancing the fundamental performance of our product. In both markets, we released new versions of our SDKs that set new standards for stability and performance, demonstrating our strong commitment to creating long-term value for our customers. Let's first talk about Agora. We recently launched our adaptive video optimization technology that can deliver exceptional live video quality and enhance the overall user experience. It leverages over 10 years of real-world experience and expertise accumulated through hundreds of billions of minutes of video usage. This advanced technology leverages various machine learning algorithms to dynamically adjust parameters and optimize performance at every step along the video processing pipeline. From the moment that video is captured to its final rendering and displaying on the viewer's screen, our technology continues to adapt to changing network conditions and device capabilities. Our adaptive video optimization technology empowers our customers to differentiate their live video applications in three key areas. Optimized image quality and matched video fluency and ultra-low latency. For example, Kumo, a social live streaming application, recently adopted our adaptive video optimization technology. Users now experience smooth, high folding video without freezing, even on older devices, and with slow internet connection. Since implementing our adaptive video optimization technology, Kumu has seen a 30% increase in session length and overall user engagement. This quarter, we also made concrete progress in engaging with Trilio customers as Studio's programmable video product continues its sunset process. Additionally, we saw increased traffic to our website and developer community, reflecting our growing mindshare among developers. I believe this will help us reach a broader spectrum of developers and customers going forward. Last week, OpenAI launched TPT 4.0, a true end-to-end multimodal text PD that can directly reason across audio, video, and text in real time. It confirms our early prediction that generative AI models will soon gain the capability to interact with human users directly in voice or video format. We anticipate a pedigree paradigm shift in the interaction between human and AI models which will inspire the next generation of killer apps. As this shift will lead to a substantial increase in the amount of voice and video feeds transmitted globally in real time, the importance of a low latency and highly reliable transmission network will be higher than ever. This will put us in a unique position to become the critical infrastructure in the AI-first future of human-computer interaction. Next, let's turn to Xuan Wang. We are thrilled to announce the launch of our new solution for live sports broadcasting. This summer, major sports events such as the European Football Championship and the Paris Olympic Games will attract billions of viewers globally. Our cloud-native solution is designed to provide customers an alternative to traditional satellite and studio-based forecasting. Compared to traditional solutions that rely on satellite and private lines, our solution uses our global network and in-house algorithms to significantly reduce latency and enhance image quality. Apart from the technical advantage, our solution also offers cost savings and greater operating flexibility for our customers. Traditionally, hosts and accommodators have to be in the same studio room or in the sports arena to cover a game. This approach incurs significant costs and also limits the number of accommodators that can work simultaneously. So most of the time, viewers could only listen to the same combinator. However, with our cloud broadcasting solution, hosts and the combinators cannot be anywhere with an internet connection. Our technology ensures that the video and audio feeds of co-combinators in the same channel are highly synchronized with the sporting events broadcasting. As a result, Many athletes, experts, and celebrities can create their own channels to cover the same game, allowing end users to choose their preferred competitors to enjoy the game. ShenWang has partnered with leading sports forecasting platforms in China to bring end users an elevated viewing experience for the upcoming European Football Championship and Paris Olympic Games this summer. I believe this new experience will trigger a major transformation in sports broadcasting, and our powerful, flexible, and cost-effective solution will become widely adopted by additional platforms to power many other live sporting events throughout the year. In 2024, we continue to host our Chaoying Su program, which strives to facilitate startups to explore and build innovative RTE applications. Over the past years, we have collaborated with re-owned VC funds, ecosystem partners, and industry leaders to bring one-stop support to startups. This year, we have brought in Moonshot AI, prominent player in foundation generative AI models to accelerate development of applications that harness the power of RTE and generative AI. Participating startups will have access to Moonshot's latest functionalities and our full portfolio of product offerings and building blocks to bring their ideas to life. We are excited to see what these innovative startups will create. The startups with the most compelling and groundbreaking applications will be showcased at our upcoming RTE conference in October. Before concluding my prepared remarks, I want to thank both Agora and Shengwang teams for their hard work and commitment during this challenging period. Let's stay focused to strengthen our technology leadership and increased market share, meanwhile moving towards sustained profitability in 2024. With that, let me turn things over to Junbo, who will review our financial results.

Thank you, Tommy. Hello, everyone. Let me start by first reviewing financial results for the first quarter of 2024, and then I will discuss outlook for the second quarter. Total revenues were $33 million in the first quarter, a decrease of 8.4% quarter-over-quarter, and a decrease of 9.4% year-over-year. Agor revenues were $15.8 million in the first quarter, an increase of 3.3% quarter-over-quarter, and an increase of 4.6% year-over-year. The increase was primarily due to the expansion and usage growth in certain verticals such as live shopping, as well as business resilience in the US market and other developed markets. Show on revenues were RMB 122.6 million in the first quarter, a decrease of 17% quarter-over-quarter, and a decrease of 16% year-over-year. The quarter-over-quarter decrease was primarily due to seasonality, since Q4 is generally the high season for digital transformation projects, and Q1 is generally the low season for social and education customers who tend to have lower usage during the Lunar New Year. The year-over-year decrease was primarily due to slowing demand from internet customers due to regulation and general economic conditions. Dollar-based net retention rate is 92% for Agora. and 78% for Shunwang, excluding revenues from discontinued business. Moving on to cost expenses. For my following comments, I will focus on non-gap adjusted financial measures, which exclude share compensation expenses, acquisition related expenses, financing related expenses, amortization expenses of acquired intangible assets, income tax related to acquired intangible assets, depreciation of property and equipment, and amortization of land-use rights. Adjusted gross margin for the first quarter was 63.2%, which was 3.9% lower than Q1 last year and 2% lower than Q4 last year. The decrease was mainly due to a change in product mix and lower utilization rate of infrastructure in Q1. Adjusted R&D expenses decreased 13.6% year-over-year to 14.6 million in Q1. Adjusted R&D expenses represented 44.2% of total revenues in the first quarter, compared to 46.3% in Q1 last year. Adjusted sales and marketing expenses or 6.3 million in Q1, decreased 25% year-over-year. Sales and marketing expenses represented 19.2% of total revenue in the quarter, compared to 23% in Q1 last year. Adjusted G&A expenses, or 6.5 million in Q1, slightly increased 6.6% year-over-year, primarily due to the increase in expected credit loss. G&A expenses represented 19.6% of total revenues in the quarter, compared to 16.8% in Q1 last year. Adjusted EBITDA was negative 6.1 million, translating to a 18.4 million, 18.4% adjusted EBITDA loss margin for this quarter, 0.7% in Q1 last year. Non-GAAP net loss was 4.8 million in Q1, translating to a 14.5% net loss margin in first quarter, significantly lower than the net loss margin of 25.1% in Q1 last year. Now turning to cash flow. Operating cash flow was negative 6.5 million in Q1, compared to negative 8.9 million last year. Free cash flow was negative 7.1 million compared to negative 9.1 million last year. Moving on the balance sheet, we ended Q1 with 380.8 million in cash, cash equivalents, bank deposits, and financial products issued by banks, or $4.13 per ADS. Net cash inflow in the quarter was mainly due to deposit received in relation to the disposal of a small portion of land for headquarter project of $19.3 million, which was offset in part by free cash flow of negative 7.1 million and share repurchase of 3.4 million. Now turning to guidance for the second quarter of 2024. We currently expect total revenues to be between $34 and $36 million. This forecast reflects our current and preliminary views on the market and operational conditions, which are subject to change. In closing, we'll continue to focus on enhancing our technology, increasing our market share, and moving toward sustained profitability in 2024. Thank you to both of Gordon and Sean's team for your hard work and contribution during this challenging period. Thank you, everyone, for attending the call today. Let's open up for questions.

Thank you. We will now begin the question and answer session. As a reminder, to ask questions, please press star 11 and wait for a name to be announced. To cancel your request, you may press star 11 again. One moment for the first question. The first question comes from the line of Yang Liu from Morgan Stanley.

Please go ahead.

Thanks for the opportunity to ask question.

Two questions here. The first one is on your customer base. We saw a sequential decline in the first quarter this year, mainly for the part. What is the reason for that? And is the customer churn behind us, or it will be an ongoing process? Yeah, that's the first question. The second one is on the outlook. As the guidance implies, some Q&Q improvement and a year-on-year stabilization of the total revenue. What is the driver? Is it mainly helped by the Agora revenue, or you are expecting The part can also recover in the second quarter. Thank you.

Thank you. I'll take both questions. So on the customer part, as I said, in Q1, revenue dropped both sequentially and year over year. And the slight decrease in number of customers was actually in line with the consistent with the decrease in revenue. to provide some color on the overall macro backdrop, the tougher regulatory environment, and also the general macroeconomic environment in China, meaning that a lot of the smaller social and entertainment startups and apps operating in much more difficult environment. And therefore, we see more turn due to customers running as a business rather than moving to other, moving to our competitors. So we expect that to continue, but hopefully that should become more moderate in the coming quarters. That's on the customer base. In terms of outlook, We actually expect both Xun1 and Agora 2 to have a sequential improvement in Q2, and actually more on the Xun1 side. And more on the Xun1 side. This is, as I explained, Q1 is generally the low season for social and also education customers. So we normally see some pickup in Q2. And also for digital transformation customers, generally Q1 is the low season due to the New Year holiday, which means the project execution will be much slower. So we see improvements on both Q1 and Q2.

Thank you. Thank you for the questions. Once again, to ask question, please press star 1-1. One moment for the next question. Our next question comes from the line of Daily Leave from Bank of America Securities. Please go ahead.

Hi, Benjamin. Thanks for taking my question. I have two questions. Firstly, regarding the AI, Benjamin mentioned there's more opportunities in this area. So, could management give some more color on the AI, you know, the incremental revenue possibility in future in which scenarios we could see more, you know, meaningful revenue in future from AI. My second question is about the expenses, the operating expenses. notice, like, there's some quote-on-quote increase for the operating expenses, especially like R&D. How do we see the trend in the following quarters for the operating expenses? Thank you.

Okay. On the AI development, it's definitely very exciting. In the previous earning calls, we actually predicted that Human users will be able to directly interact with AI models in voice and video formats. Initially, it's text-based, but we consider it will happen in voice. What happens is actually faster than we imagined. The launch of GPT-4.0 is ahead of our expectations. Thanks to the rapid evolution and global arm-raise of generative AI models, With generating AI models, multi-modal capabilities, there will be another dimension being added to RT activities. RT activities will expand from human to human and human to machine or device to also include human to generating AI models. of real-time engagement will stand by that. In the long run, this will hugely increase the amount of RT activities and enrich people's lives. In the next few years, we'll be able to see many more use cases emerge and mature, such as AI-based interactive education, customer service, personal assistant, social and gaming use cases, voice and video conversations will become the new norm of interaction between human users and cloud agents powered by generating AI models. As a result, massive amount of real-time voice and video will flow through global internet, and latency will become a critical factor. When introducing GPT-4.0, OpenAI actually mentioned it has an average response time of 320 milliseconds. However, when an overseas user talks with a model, round-trip transmission latency needs to be added on top of the model's response latency, and the experience could change from exceptional to unbearable. To deliver a low latency, highly reliable, and high-quality user experience, Our technology is the right choice for foundation model AI companies and for companies who build applications on top of generative AI models.

Thank you, Tony. So I'll take a question on the operating expenses. So I think in our last earnings call, we already explained that the low operating expenses was obviously due to cost control measures. But at the same time, it was also It was below a normal level, and due to reversal of certain accrued expenses, we accrued year-end bonus and performance bonus every quarter. However, given the top operating environment last year and the final operating result was below our internal budget, so the actual realized year-end bonus and performance bonus were lower than Z. So there was some reversal in Q4, which caused the Q4 quality operating expenses to be abnormally low. And that's why there was some small sequential increase from Q4 to Q1. So looking forward this year, we are still very cautious about the overall operating environment. So we will continue to manage our expenses very carefully. And we do not expect OPEX in general, and including R&D expenses, to increase sequentially from Q1 onward. And if anything, we will try to control the overall expenses a little bit more.

Thank you, Tony. Thank you. Thank you.

Thank you for the questions. Once again, to ask questions, please press star 11 and wait for your name to be announced. There are no further questions. Thank you, everybody, for attending the company's call today. As a reminder, the card recording in the earnings release will be available on the company's website at investor.agora.io. And if there are any questions, please feel free to email the company. Thank you. That concludes today's conference call. Thank you for your participating. You may now disconnect.

Thank you. Thank you. Thank you.

Thank you.

Good day and thank you for standing by. Welcome to Agora Inc's first quarter 2024 financial results conference call. At this time, all participants are in a listen-only mode. After the speaker's presentation, there will be a question and answer session. To ask a question during the session, you need to press star 11 on your telephone. You'll then hear an automated message advising your hand is raised. Please be advised that today's conference is being recorded. The company's earnings results press release, earnings presentations, SEC filing and a replay of today's call can be found on its IR website at investor.agora.io. Joining me today are Tony Chow, founder, chairman and CEO, Ching-Po Wang, the company's CFO. Reconciliations between the company's GAAP and non-GAAP results can be found in its earnings press release. During this call, the company will make forward-looking statements about its future financial performance and other future events and trends. These statements are only predictions that are based on what the company believes today and actual results may differ materially. These forward-looking statements are subject to risk, uncertainties, assumptions and other factors that could affect the company's financial results and the performance. of its business and which the company discussed in detail in its filing with the SEC, including today's earnings press release and the risk factors and other information contained in the final prospectus relating to its initial public offering. Agora Inc. remains no obligation to update any forward-looking statements the company may make on today's call. With that, let me turn it over to Tony. Hi, Tony.

Thanks, Peter. and welcome everyone to our earnings call. I'll first reveal our operating result in the past quarter. Agora revenues were $15.8 million in the first quarter, up 3% quarter over quarter. This is a great result considering the macroeconomic challenge under the high interest rate environment and is mainly driven by usage growth from emerging use cases such as live shopping, As of the end of this quarter, Agora had over 1,700 active customers, up 16% compared to one year ago. Xiong Wang's revenues were 122.6 million RMB in the first quarter, down 16% year-over-year, mainly due to challenging macroeconomic and regulatory environment. as well as the disposal of the customer engagement cloud business in the first quarter of 2023. As of the end of this quarter, Shunwang had over 3800 of active customers, down 2% compared to one year ago. Now moving on to our business, product and technology updates for this quarter. Despite a challenging operating environment, we continued to focus on enhancing the fundamental performance of our products. In both markets, we released new versions of our SDKs that set new standards for stability and performance, demonstrating our strong commitment to creating long-term value for our customers. Let's first talk about Agora. We recently launched our adaptive video optimization technology that can deliver exceptional live video quality and enhance the overall user experience. It leverages over 10 years of real-world experience and expertise accumulated through hundreds of billions of minutes of video usage. This advanced technology leverages various machine learning algorithms to dynamically adjust parameters and optimize performance at every step along the video processing pipeline. From the moment that video is captured to its final rendering and display on the viewer's screen, our technology continues to adapt to changing network conditions and device capabilities. Our adaptive video optimization technology empowers our customers to differentiate their live video applications in three key areas. Optimized image quality and matched video fluency and ultra-low latency. For example, Kumo, a social live streaming application, recently adopted our adaptive video optimization technology. Users now experience smooth, high folding video without freezing, even on older devices, and with slow internet connection. Since implementing our adaptive video optimization technology, Kumu has seen a 30% increase in session length and overall user engagement. This quarter, we also made concrete progress in engaging with Trilio customers as Studio's programmable video product continues its sunset process. Additionally, we saw increased traffic to our website and developer community, reflecting our growing mindshare among developers. I believe this will help us reach a broader spectrum of developers and customers going forward. Last week, OpenAI launched TPT 4.0, a true end-to-end multimodal text PD that can directly reason across audio, video, and text in real time. It confirms our early prediction that generative AI models will soon gain the capability to interact with human users directly in voice or video format. We anticipate a pedigree paradigm shift in the interaction between human and AI models which will inspire the next generation of killer apps. As this shift will lead to a substantial increase in the amount of voice and video feeds transmitted globally in real time, the importance of a low latency and highly reliable transmission network will be higher than ever. This will put us in a unique position to become the critical infrastructure in the AI-first future of human-computer interaction. Next, let's turn to Shen Wang. We are thrilled to announce the launch of our new solution for live sports broadcasting. This summer, major sports events such as the European Football Championship and the Paris Olympic Games will attract billions of viewers globally. Our cloud-native solution is designed to provide customers an alternative to traditional satellite and studio-based broadcasting. Compared to traditional solutions that rely on satellite and private lines, our solution uses our global network and in-house algorithms to significantly reduce latency and enhance image quality Apart from the technical advantage, our solution also offers cost savings and greater operating flexibility for our customers. Traditionally, hosts and accommodators have to be in the same studio room or in the sports arena to cover a game. This approach incurs significant costs and also limits the number of accommodators that can work simultaneously. So most of the time, viewers could only listen to the same combinator. However, with our cloud broadcasting solution, hosts and the combinators cannot be anywhere with an internet connection. Our technology ensures that the video and audio feeds of co-combinators in the same channel are highly synchronized with the sporting events broadcasting. As a result, Many athletes, experts, and celebrities can create their own channels to cover the same game, allowing end users to choose their preferred competitors to enjoy the game. ShenWang has partnered with leading sports forecasting platforms in China to bring end users an elevated viewing experience for the upcoming European Football Championship and Paris Olympic Games this summer. I believe this new experience will trigger a major transformation in sports broadcasting, and our powerful, flexible, and cost-effective solution will become widely adopted by additional platforms to power many other live sporting events throughout the year. In 2024, we continue to host our Chaoying Su program, which strives to facilitate startups to explore and build innovative RTE applications. Over the past years, we have collaborated with re-owned VC funds, ecosystem partners, and industry leaders to bring one-stop support to startups. This year, we have brought in Moonshot AI, prominent player in foundation generative AI models to accelerate development of applications that harness the power of RTE and generative AI. Participating startups will have access to Moonshot's latest functionalities and our full portfolio of product offerings and building blocks to bring their ideas to life. We are excited to see what these innovative startups will create. The startups with the most compelling and groundbreaking applications will be showcased at our upcoming RTE conference in October. Before concluding my prepared remarks, I want to thank both Agora and Shengwang teams for their hard work and commitment during this challenging period. Let's stay focused to strengthen our technology leadership and increased market share, meanwhile moving towards sustained profitability in 2024. With that, let me turn things over to Jingbo, who will review our financial results.

Thank you, Tommy. Hello, everyone. Let me start by first reviewing financial results for the first quarter of 2024, and then I will discuss outlook for the second quarter. Total revenues were $33 million in the first quarter, a decrease of 8.4% quarter-over-quarter, and a decrease of 9.4% year-over-year. Agor revenues were $15.8 million in the first quarter, an increase of 3.3% quarter-over-quarter, and an increase of 4.6% year-over-year. The increase was primarily due to the expansion and usage growth in certain verticals such as live shopping, as well as business resilience in the US market and other developed markets. Show on revenues were RMB 122.6 million in the first quarter, a decrease of 17% quarter-over-quarter, and a decrease of 16% year-over-year. The quarter-over-quarter decrease was primarily due to seasonality, since Q4 is generally the high season for digital transformation projects, and Q1 is generally the low season for social and education customers who tend to have lower usage during the Lunar New Year. The year-over-year decrease was primarily due to slowing demand from internet customers due to regulation and general economic conditions. Dollar-based net rotation rate is 92% for Agora, and 78% , excluding revenues from discontinued business. Moving on to cost expenses. For my following comments, I will focus on non-gap adjusted financial measures, which exclude compensation expenses, acquisition related expenses, financing related expenses, amortization expenses of acquired intangible assets, income tax related to acquired intangible assets, depreciation of property and equipment, and amortization of land use rights. Adjusted gross margin for the first quarter was 63.2%, which was 3.9% lower than Q1 last year and 2% lower than Q4 last year. The decrease were mainly due to a change in product mix and lower utilization rate of infrastructure in Q1. Adjusted R&D expenses decreased 13.6% year-over-year to 14.6 million in Q1. Adjusted R&D expenses represented 44.2% of total revenues in the first quarter, compared to 46.3% in Q1 last year. Adjusted sales and marketing expenses or 6.3 million in Q1, decreased 25% year-over-year. Sales and marketing expenses represented 19.2% of total revenue in the quarter, compared to 23% in Q1 last year. Adjusted G&A expenses, or 6.5 million in Q1, slightly increased 6.6% year-over-year, primarily due to the increase in expected credit loss. G&A expenses represented 19.6% of total revenues in the quarter, compared to 16.8% in Q1 last year. Adjusted EBITDA was negative 6.1 million, translating to a 18.4% adjusted EBITDA loss margin both quarter, compared to 17.4%. 0.7% in Q1 last year. Non-GAAP net loss was 4.8 million in Q1, translating to a 14.5% net loss margin fourth quarter, significantly lower than the net loss margin of 25.1% in Q1 last year. Now turning to cash flow. Operating cash flow was negative 6.5 million in Q1, compared to negative 8.9 million last year. Free cash flow was negative 7.1 million compared to negative 9.1 million last year. Moving on the balance sheet, we ended Q1 with 380.8 million in cash, cash equivalents, bank deposits, and financial products issued by banks, or $4.13 per ADS. Net cash inflow in the quarter was mainly due to deposit received in relation to the disposal of a small portion of land for headquarter project of $19.3 million, which was offset in part by free cash flow of negative 7.1 million and share repurchase of 3.4 million. Now turning to guidance for the second quarter of 2024. We currently expect total revenues to be between $34 and $36 million. This forecast reflects our current and preliminary views on the market and operational conditions, which are subject to change. In closing, we'll continue to focus on enhancing our technology, increasing our market share, and moving toward sustained profitability in 2024. Thank you to both of Gordon and Sean's team for your hard work and contribution during this challenging period. Thank you everyone for attending the call today. Let's open up for questions.

Thank you. We will now begin the question and answer session. As a reminder, to ask questions, please press star 1 1 and wait for a name to be announced. To cancel your request, you may press star 1 1 again. One moment for the first question. The first question comes from the line of Yang Liu from Morgan Stanley.

Please go ahead.

Thanks for the opportunity to ask question.

Two questions here. The first one is on your customer base. We saw a sequential decline in the first quarter this year, mainly for the part. What is the reason for that? And is the customer churn behind us, or it will be an ongoing process? Yeah, that's the first question. The second one is on the outlook. As the guidance implies, some Q&Q improvement on the year-on-year stabilization of the total revenue. What is the driver? Is it mainly helped by the agora of revenue, or you are expecting the part can also recover in the second quarter. Thank you.

Thank you. I'll take both questions. So on the customer part, as I said, in Q1, revenue dropped both sequentially and year over year. And the slight decrease in number of customers was actually in line with the consistent with the decrease in revenue. to provide some color on the overall macro backdrop. The tougher regulatory environment and also the generic macroeconomic environment in China means that a lot of the smaller social and entertainment startups and apps operating in a much more difficult environment. And therefore, we see more churn due to customers running out of business rather than moving to our competitors. So we expect that to continue, but hopefully that should become more moderate in the coming quarters. So that's on the customer base. In terms of the outlook, We actually expect both Xun1 and Agora 2 to have a sequential improvement in Q2, and actually more on the Xun1 side. And more on the Xun1 side. This is, as I explained, Q1 is generally the low season for social and also education customers. So we normally see some pickup in Q2. And also for digital transformation customers, generally Q1 is the low season due to the New Year holiday, which means the project execution will be much slower. So we see improvements on both Q1 and Q2.

Thank you. Thank you for the questions. Once again, to ask question, please press star 1-1. One moment for the next question. Our next question comes from the line of David Lee from Bank of America Securities. Please go ahead.

Hi, Benjamin. Thanks for taking my question. I have two questions. Firstly, regarding the AI, Benjamin mentioned there's more opportunity more opportunities in this area so uh commencement gives some more color on the ai you know the incremental revenue possibility in future uh in voyage scenarios uh we could see more you know meaningful revenue in future from ai uh my second question is about the um expenses the operating expenses so we notice, like, there's some quote-on-quote increase for the operating expenses, especially like R&D. How do we see the trend in the following quarters for the operating expenses? Thank you.

Okay. On the AI development, it's definitely very exciting. In the previous earning calls, we actually predicted that Human users will be able to interact with AI models in voice and video formats. Initially, it's text-based, but we consider it will happen in voice. What happens is actually faster than we imagined. The launch of GPT-4.0 is ahead of our expectation. Thanks to the rapid evolution and global arm race of generative AI models, With generating AI models, multi-modal capabilities, there will be another dimension being added to RTE activities. RTE activities will expand from human to human and human to machine or device to also include human to generating AI models. of real-time engagement will extend by that. In the long run, this will hugely increase the amount of RT activities and enrich people's lives. In the next few years, we'll be able to see many more use cases emerge and mature, such as AI-based interactive education, customer service, personal assistant, social and gaming use cases, voice and video conversations will become the new norm of interaction between human users and cloud agents powered by generating AI models. As a result, massive amount of real-time voice and video will flow through global internet, and latency will become a critical factor. When introducing GPT-4.0, OpenAI actually mentioned it has an average response time of 320 milliseconds. However, when an overseas user talks with a model, round-trip transmission latency needs to be added on top of a model's response latency, and the experience could change from exceptional to unbearable. To deliver a low latency, highly reliable, and high-quality user experience, Our technology is the right choice for foundation model AI companies and for companies who build applications on top of generative AI models.

Thank you, Tony. So I'll take a question on the operating expenses. So I think in our last earnings call, we already explained that the low operating expenses was obviously due to our cost control measures. But at the same time, it was also It was below a normal level and due to reversal of certain accrued expenses. We accrued year-end bonus and performance bonus every quarter. However, given the top operating environment last year and the final operating result was below our internal budget, so the actual realized year-end bonus and performance bonus were lower than the earlier accruals, so there was some reversal in Q4, which caused the Q4 quarterly operating expenses to be abnormally low, and that's why there was some small sequential increase from Q4 to Q1. So looking forward this year, we are still very cautious about the overall operating environment, so we will continue to manage our expenses very carefully And we do not expect OPEX in general, and including R&D expenses, to increase sequentially from Q1 onward. And if anything, we will try to control the overall expenses . Thank you, Tony.

Thank you.

Thank you for the questions. Once again, to ask questions, please press star 11 and wait for your name to be announced. There are no further questions. Thank you, everybody, for attending the company's call today. As a reminder, the card recording in the earnings release will be available on the company's website at investor.agora.io. And if there are any questions, please feel free to email the company. Thank you. That concludes today's conference call. Thank you for your participating. You may now disconnect.