The paragraph is a transcript of the opening remarks from an AMD conference call for their fourth quarter and full year 2024 financial results. The call is introduced by the operator, who explains that participants are in listen-only mode until a question-and-answer session. Matt Ramsey, Vice President of Investor Relations, welcomes attendees and notes the availability of earnings materials on AMD's website. He highlights that the call will focus primarily on non-GAAP financial measures, with reconciliations provided in the press release and slides. Participants include Dr. Lisa Su, CEO, and Jean Hu, CFO. Ramsey notes the inclusion of forward-looking statements and hands the call over to Lisa Su, who describes 2024 as a transformative year for AMD.

The company achieved significant growth in its data center AI franchise, with the data center segment accounting for about 50% of annual revenue. Driven by increased adoption of Instinct and EPYC processors by cloud, enterprise, and supercomputing customers, fourth-quarter revenues hit a record $7.7 billion, an increase of 24% year-over-year. On an annual basis, revenue grew 14% to $25.8 billion, with the data center's revenue nearly doubling and the client segment growing 52%, offsetting declines in Gaming and Embedded segments. Notably, the data center segment's revenue increased 69% year-over-year to $3.9 billion. The company gained significant market share in cloud computing, exiting 2024 with over 50% share among major hyperscale customers, as demand for EPYC CPUs surged. AWS, Alibaba, Google, Microsoft, and Tencent launched numerous AMD instances in the fourth quarter, highlighting the strong market adoption.

In 2024, AMD experienced significant growth in the adoption of its EPYC processors, both in cloud instances and on-premise enterprise sales, with notable deployments by major companies like Akamai, Verizon, and Visa. The number of available EPYC platforms increased to over 450, with new offerings from major server manufacturers, while the Turin processors set over 540 performance records. Looking ahead to 2025, AMD anticipates further growth in both cloud and enterprise markets due to its robust processor portfolio. The company also saw notable success in its data center AI business, generating over $5 billion in revenue, thanks to advancements in its Instinct accelerators and ROCm software suite.

In the fourth quarter, AMD's MI300X production deployments expanded with major cloud partners such as Meta, Microsoft, and IBM, who are using them for AI models and services. The MI300X is being deployed across various cloud service providers (CSPs) and integrated into enterprise solutions by Dell. AMD's adoption in high-performance computing (HPC) is also significant, powering some of the fastest and most energy-efficient supercomputers, including the El Capitan system at Lawrence Livermore National Labs, which is the world's fastest supercomputer. The University of Stuttgart's Hunter supercomputer also uses AMD's MI300A.

The paragraph highlights AMD's advancements in AI and scientific research, particularly through the use of their ROCm software stack. In 2024, AMD aims to position ROCm as a leading open AI software stack, enhancing developer options and accelerating innovation. Over a million models on Hugging Face now run on AMD platforms with support for major frameworks, and the company has expanded large-scale ROCm deployments. Following performance improvements, including ROCm 6.3's release, AMD's MI300X inferencing performance has markedly improved. Ongoing efforts include biweekly container updates and increased open-source community involvement. Additionally, volume production of the MI325X product began in the fourth quarter, progressing well to support new customers.

The paragraph discusses the strong market position and performance advantages of AMD's MI325 and future MI350 series, highlighting recent large customer wins and the generational leap in AI performance expected with the upcoming CDNA 4 architecture. The MI350 series is receiving positive feedback and strong interest, with plans to send samples to lead customers soon and production shipments expected by mid-year. Additionally, development is progressing well on the MI400 series, planned for launch in 2026, which aims to enable powerful data center-scale solutions with integrated networking, CPU, and GPU capabilities.

The paragraph discusses the company's acquisition of ZTE Systems, highlighting key milestones like receiving regulatory approvals in several jurisdictions and positive customer feedback, particularly in cloud and OEM markets. The company plans to divest ZTE's U.S.-based data center production capabilities soon after the acquisition is finalized. In the client segment, the company achieved a 58% year-over-year revenue increase to a record $2.3 billion, driven by strong demand for Ryzen desktop and mobile processors. Ryzen dominated CPU sales in various global retail channels during the holiday period and achieved record OEM PC sell-through shares. A new collaboration with Dell will bring a full portfolio of Ryzen Pro-powered commercial PCs, with an expected launch this spring. At CES, the company unveiled 22 new mobile Ryzen processors, enhancing its leadership in compute, graphics, and AI capabilities.

The article discusses the company's performance and future plans across various segments. In the AI PC segment, the company highlights its comprehensive CPU portfolio, which supports Windows Copilot experiences, and anticipates mid-single-digit growth in the PC TAM by 2025. The company expects to outpace market growth due to its strong client CPU offerings. The Gaming segment saw a 59% revenue decline, attributed to reduced channel inventory by Microsoft and Sony, though future normalization is expected by 2025. In Gaming Graphics, revenue decreased due to preparations for the Radeon 9000 series GPU launch, focusing on enthusiast markets with enhanced performance features. The Embedded segment experienced a 13% revenue drop, with mixed demand reflecting strength in aerospace and defense but weaknesses in industrial and communications markets. The company continues to expand its adaptive computing portfolio with innovative solutions.

The paragraph outlines key achievements and future outlook for the company. In 2024, they launched the Versal RF series for aerospace and defense and introduced the Versal premium series Gen 2 with advanced compute capabilities. They shipped a next-gen Alveo card for ultra-low latency trading and gained adaptive computing market share, securing a record $14 billion in design wins. With record revenue for both the quarter and year, EPYC and Ryzen processor gains strengthened their CPU portfolio. Their data center AI business, along with Instinct hardware and ROCm software, advanced significantly. Looking to 2025, they anticipate strong growth in data center and client businesses and modest growth in gaming and embedded sectors, forecasting double-digit revenue and EPS growth. The paragraph concludes by noting rapid AI innovation driven by investments and breakthroughs from notable AI initiatives.

The paragraph discusses AMD's advancements and growth initiatives in AI, emphasizing their investment in compute capacity and software innovation. It highlights AMD's unique position to provide comprehensive AI solutions across various devices and its progress in securing a significant market share. AMD anticipates substantial long-term growth, particularly in its data center AI segment, projecting revenue growth from $5 billion in 2024 to tens of billions in the coming years. Jean Hu then reviews AMD's financial performance, noting record revenue of $25.8 billion for 2024, a 14% increase driven by data center and client segment growth, gross margin expansion, and a focus on AI investments for future growth.

The company's revenue grew by 12% sequentially, largely due to the client, data center, and gaming segments, with a gross margin increase to 54%. Operating expenses rose 23% to support R&D and marketing, contributing to a new record operating income of $2 billion and a 26% operating margin. The data center segment saw a revenue record of $3.9 billion, up 69% year-over-year, driven by AMD Instinct GPU and EPYC CPU sales. The client segment also reached a record $2.3 billion, a 58% increase due to demand for Ryzen processors. The gaming segment decreased by 59% to $563 million, largely due to declining semi-customer revenue, while the embedded segment fell 13% to $923 million due to varied market demand.

In the recent quarter, the company generated $1.3 billion in operational cash and $1.1 billion in free cash flow, with inventory increasing to $5.7 billion. The company ended the quarter with $5.1 billion in cash and investments. They repurchased 1.8 million shares, returning $256 million to shareholders, and for the year, repurchased 5.9 million shares, returning $862 million. They have $4.7 billion left for share repurchase. Starting fiscal year 2025, they plan to combine the client and gaming segments into one reportable segment while continuing distinct revenue disclosures. They project first-quarter 2025 revenue of around $7.1 billion, up 30% year-over-year, led by data center and client growth, despite declines in gaming and embedded businesses. Revenue is expected to decrease 7% sequentially. They anticipate a 54% non-GAAP gross margin, $2.1 billion in non-GAAP operating expenses, $24 million in non-GAAP net income, a 13% tax rate, and a diluted share count of about 1.64 billion. 2024 was a strong year for AMD, marked by revenue growth and increased earnings.

The paragraph discusses the progress and future expectations for the GPU business focusing on 2025. Jean mentions a plan to capitalize on momentum for revenue growth, and Lisa Su highlights the successful execution of the data center GPU business in 2024. She expresses satisfaction with the progress in both hardware and software roadmaps. New designs, particularly the MI325 and MI300, were initiated at the end of 2024 and will be deployed in the first half of 2025. Anticipation is also building for the MI350 series, indicating ongoing development and potential growth in the business.

The paragraph discusses the company's plans to accelerate the production ramp of a product due to strong customer demand, which will improve competitiveness. This is expected to drive significant growth in the data center business, particularly in the server and data center GPU product lines, with strong double-digit growth anticipated for 2024 and 2025. The MI350 will be a key catalyst for this growth. In terms of financial guidance, the company predicts a 7% sequential downturn for Q1, with the data center segment aligning with this average. The client and embedded businesses are expected to decline more due to seasonality, while the gaming business will see a smaller decrease as it recovers from inventory normalization.

In the paragraph, Timothy Arcuri from UBS asks about the server CPU business and whether competition is affecting pricing strategies and growth expectations. Jean Hu responds confidently, stating their company has a strong lineup with products like Turin, Genoa, and Milan, offering the best total cost of ownership, and is optimistic about continuing growth in the server CPU market from both unit and average selling price perspectives. Timothy also inquires about data center GPU performance for December, estimating it to be around $2 billion, and asks about expectations for March. Jean Hu notes that their data center business performed well, with the server segment outperforming data center GPUs, yet overall results were consistent with expectations.

In the paragraph, Vivek Arya from Bank of America Securities questions AMD's CEO, Lisa Su, about the company's approach to forecasts for their data center GPU business. Whereas last year AMD set specific targets, this year they have not. Lisa Su explains that with the business having scaled to over $5 billion, future guidance will focus more on the overall segment with qualitative insights into the specifics of each business line. She confirms confidence in strong demand for AI compute, highlighting the potential for the data center market to reach $500 billion by 2028. Without providing a specific forecast for 2025, she suggests the business is expected to grow into the tens of billions within a few years.

The paragraph discusses the advancements and confidence in a business's product roadmap, highlighting the progression from the MI300 to the MI400 series, which is expected to bring significant innovation and excitement. The speaker, Lisa Su, responds to a question from Vivek Arya about the implications of recent announcements from DeepSeek and other organizations, noting that these innovations reflect the rapid pace of AI development. Su emphasizes that advancements in AI models and algorithms, which require less infrastructure, are beneficial for AI adoption and that the business supports open-source initiatives.

The paragraph discusses the optimistic outlook on the adoption of open-source AI models and infrastructure investments with OpenAI and Stargate, indicating a rapid innovation phase in AI. It emphasizes the need for diverse computing resources—CPUs, GPUs, ASICs, and FPGAs—due to the varied AI workloads. While ASICs are regarded as important, their role is seen as smaller compared to GPUs due to programmability and adaptability to changing AI algorithms. The AMD portfolio encompasses these computing types to serve as a comprehensive compute partner. The paragraph also notes an interaction with Joshua Buchalter from TD Cowen, mentioning MI355X's timeline.

In this conversation, Lisa Su discusses the outlook for AMD's data center GPUs, indicating a positive ramp in 2025. The MI350 series, initially slated for the second half, is now expected midyear, seen as a positive change. New AI design wins with MI300 and MI325 are anticipated in the first half of the year, and the MI350 will contribute with higher ASPs and larger deployments. Regarding their client business, Lisa Su notes a strong performance throughout 2024, especially in Q4, driven by certain factors, and clarifies that there is no significant inventory buildup despite the outperformance in the broader PC market during the second half of the year.

The paragraph discusses the strong adoption and performance of new products, particularly in the desktop and notebook markets. The company experienced its highest desktop sell-out in years during the holiday season with the launch of new gaming CPUs, which faced market constraints but had strong continued shipping. On the notebook side, new AI PCs and Copilot+ PC compatible products launched by OEM partners also performed well. There were no significant impacts from tariffs in Q4. The company anticipates typical seasonal trends in Q1, with the desktop business performing better than expected and the mobile business showing typical seasonality. The company is optimistic about growth prospects for 2025 due to a strong product portfolio and market dynamics. An operator then introduces Harlan Sur from JPMorgan for a question about the server CPU business growth.

The paragraph discusses AMD's strong growth in its server business, particularly in cloud and enterprise sectors, with a sequential double-digit increase in the fourth quarter. Lisa Su highlights the successful performance of AMD's server business, which is bolstered by investments in go-to-market strategies and a robust portfolio. These investments have led to significant new proof-of-concepts converting into volume deployments. Looking ahead to 2025, AMD remains competitive across various price points and workloads. The discussion also touches on AMD's role in AI infrastructure, emphasizing efforts in networking through technologies like Ethernet and Infinity Fabric, dispelling any misconceptions about falling behind in this area.

In the paragraph, Lisa Su discusses the importance of networking in AI solutions and outlines AMD's efforts in this area. She mentions that AMD has been actively investing and collaborating with partners to improve networking performance, particularly as they progress from the MI300 to MI400 series. Su highlights their work on Ethernet technology and the development of an in-house AI Network Interface Card (NIC) with the Pensando team. Additionally, she notes that AMD is implementing full rack solutions in their MI350 and MI400 series, emphasizing the significance of system-level scaling. Lastly, Blayne Curtis raises a question about the expected slower sequential growth of AMD's data center GPU business in the upcoming quarters.

In the paragraph, Lisa Su discusses AMD's progress and strategy in the data center GPU market. She highlights the significant growth and expansion with new customer deployments, particularly focusing on the transition to the MI350 series GPUs. Su mentions that although there is no major shift in conversations about ASICs, AMD aims to strengthen its position in both inference and training solutions. She emphasizes the company's step-wise roadmap, significant software investments, and the importance of scaling up GPU deployments in data centers. Overall, AMD is making good progress in addressing customer demands and expanding its product offerings.

The paragraph discusses AMD's involvement in ASICs, highlighting their significance as part of AMD's offerings alongside semi-custom solutions. AMD has been engaging in discussions with customers about using their IP for innovation. The paragraph also features a Q&A section where Stacy Rasgon from Bernstein Research questions AMD's server and GPU business performance. Responding, Lisa Su clarifies a misunderstanding about the growth figures, emphasizing that while the data center segment increased by 9% sequentially, the server and data center GPU growth rates varied slightly. She suggests that some external models might underestimate their Q3 data center GPU performance.

In the paragraph, there is a discussion about the outlook for the server CPU market in 2025. Lisa Su mentions the expectation of a healthy server market, with a focus on both cloud and enterprise segments. She observes that in early 2024, investments in CPUs decreased as companies prioritized AI, but these investments began to pick up again in the second half of the year, with expectations for this trend to continue into 2025. She notes that enterprise refresh cycles are resuming, and several cloud vendors are updating their data centers. Additionally, there is a brief mention of competitive intensity concerning custom silicon, though details on this aspect are limited in the provided excerpt.

In the paragraph, Lisa Su discusses AMD's approach to the server market, emphasizing the importance of having the right product for specific workloads rather than focusing solely on custom silicon versus x86. She highlights AMD's competitive strategy through their Zen 4 and Zen 5 product lines, which cater to both cloud native and enterprise needs by offering a range of design points and performance metrics. Su expresses confidence in AMD's growing market share due to their strong product portfolio and trusted delivery capabilities. In the follow-up, Jean Hu addresses AMD's gross margin guidance of 54% for the first quarter, noting that it's primarily driven by revenue mix. She mentions that both the data center and client businesses are growing year-over-year, and the revenue mix is consistent with the previous quarter. This implies an expectation for margin improvement in the second half of the year.

The paragraph discusses AMD's financial outlook, stating that the company expects its gross margin to remain consistent at 54% in the first half due to the current revenue mix. However, in the second half, the company anticipates growth driven by its data center segment, which is expected to increase the gross margin. The call concludes with appreciation expressed by Matt Ramsey and the operator to the participants of the teleconference.

This summary was generated with AI and may contain some inaccuracies.