The paragraph is a transcript of the opening remarks of the AMD First Quarter 2025 Conference Call. The operator introduces Matt Ramsey, the Vice President of Financial Strategy and Investor Relations, who begins by welcoming participants and referring them to the earnings press release and slides available on AMD's website. He notes that the call will focus on non-GAAP financial measures, with reconciliations provided online. Participants on the call include Dr. Lisa Su, Chair and CEO, and Jean Hu, CFO and Treasurer. The call is live and will be available for replay. Ramsey mentions upcoming conferences featuring AMD executives and provides a caution about forward-looking statements and associated risks.

The paragraph outlines AMD's strong start to the year, highlighting a 36% year-over-year revenue increase to $7.4 billion, driven by robust growth in their data center and client and gaming segments. The company achieved higher revenue and earnings per share than expected, with significant sales growth in Instinct AI Accelerator, Epic, and Ryzen CPUs. The data center segment alone saw a 57% revenue increase due to the success of fourth and fifth-gen EPYC processors, with major cloud providers expanding their deployments. Despite challenging macroeconomic conditions, AMD’s diverse product portfolio and execution underscore their readiness for continued growth into 2025.

AWS introduced new FPGA accelerated instances using EPYC processors and Xilinx Virtex FPGAs, which are optimized for demanding tasks like genomics and cloud broadcasting. Major cloud providers are advancing Turin programs with a focus on EPYC adoption, leading to a significant year-over-year increase in instances activated by Forbes 2000 companies across various sectors. Notable updates include CrowdStrike's efficiency gains with EPYC, Siemens' software-defined vehicle solution utilizing Azure's EPYC CPUs and Radeon Pro GPUs, and Oracle's enhanced Exadata X11m platform, which offers improved performance with fifth-gen EPYC processors. On-premises, EPYC CPU sales have continued strong growth, especially in key industries like automotive, finance, and technology.

The paragraph highlights the growing momentum and deployment of EPYC-based platforms across various top industries, such as telecom, aerospace, and semiconductor sectors, and anticipates accelerating enterprise adoption with the availability of new Turin platforms. Plans include the manufacturing of fifth-gen EPYC processors at TSMC's Arizona facility, with production starting in 2025, and the future launch of Venice processors using TSMC's 2-nanometer process in 2026. Additionally, the paragraph notes significant revenue growth in the data center AI sector, with increased shipments and deployments of MI325X and MI300 series platforms, supporting various AI applications with expanded use by hyperscalers.

In the recent quarter, AMD expanded its customer base by adding major cloud, enterprise, and frontier model developers using its Instinct GPUs for inferencing and pre-training AI workloads. The company is scaling deployments from single nodes to distributed clusters, with significant growth in training engagements for hyperscale, AI, and enterprise customers. AMD is also advancing Sovereign AI projects, including a partnership with G42 to establish a major AI compute facility in France. On the software side, AMD accelerated its update schedule for ROCm, offering bi-weekly releases with performance optimizations, and enhanced its open-source community support, enabling over two million models on Hugging Face to operate seamlessly on AMD hardware.

The paragraph discusses the company's efforts to enhance its Instinct accelerators by ensuring new models, like Meta's LLAMA 4 and Google's GEMMA 3, launch with immediate support. Post-launch, they provide regular updates like ROCm 6.4 to boost performance across various AI frameworks. The company completed its acquisition of ZT Systems to enhance its AI solutions, enabling faster deployment for hyperscalers and OEM partners, with plans to optimize upcoming MI400 and MI350 series designs. The MI350 series is based on the advanced CDNA 4 Architecture and is set for production mid-year, with strong performance indications and ongoing customer collaborations.

The paragraph discusses AMD's advancements in AI and client segments. The new CDNA 4 is designed to deliver significant performance improvements across AI workloads, surpassing previous iterations in memory capacity and throughput. The MI350 series is generating strong customer interest with upcoming large-scale deployments, such as a partnership with Oracle. AMD's development of the MI400 series is on track for next year, aimed at enhancing AI infrastructure scalability. In parallel, AMD's client and gaming segment saw a 28% revenue increase, driven by a 68% rise in client revenue, boosted by high-end Ryzen processors and increased desktop sales. The company plans to reveal further details at an AI event in June.

AMD experienced strong sales and demand for its latest generation Ryzen processors, particularly in the gaming segment, as the Ryzen 9 9950 X3D and new AI-enhanced mobile processors received positive reviews. Notebook sales, including those with Ryzen AI capabilities, saw a significant increase, establishing AMD as a preferred choice for gaming, ultra-thin, and commercial laptops. The company's commercial PC segment grew, with notable increases in systems powered by AMD from major brands and new business wins across various industries. However, gaming revenue decreased by 30% year-over-year due to lower semi-custom sales, despite higher Radeon graphics sales. Looking forward, AMD is optimistic about growing its client processor revenue ahead of the market by 2025, driven by expanding its desktop and notebook portfolio. Console channel inventories have normalized, with stronger demand signals for 2025.

The paragraph highlights the strong demand and successful launch of the Radeon 9070 series, powered by the RDNA 4 Architecture, and the introduction of FSR 4 technology, enhancing gaming experiences. Despite a slight decrease in first-quarter revenue for the embedded segment, there's optimism for growth in 2025 driven by improving demand in specific markets and new product launches, such as the Spartan UltraScale Plus FPGAs and EPYC-embedded 9005 series CPUs. Notable collaborations include Cisco and IBM utilizing these CPUs for networking and storage solutions. The release of the Vitis AI software suite aims to expand edge AI applications. Overall, the positive first quarter results and second quarter projections indicate momentum across the business.

The paragraph discusses AMD's outlook and strategy in the face of dynamic macro and regulatory challenges, including recent export controls affecting shipments to China. Despite these challenges, AMD anticipates strong revenue growth in 2025, driven by their Zen 5 EPYC and Ryzen CPUs, Radeon GPUs, and Instinct MI350 accelerators. They expect growth in the semi-custom and embedded businesses due to reduced inventory and improved demand. AMD is investing in technology, market initiatives, and AI solutions to capitalize on growth opportunities and differentiate itself in the market. Lisa then hands over to Jean Hu for a financial review and second-quarter outlook for fiscal 2025.

In the first quarter of fiscal year 2025, the company reported record revenue of $7.4 billion, a 36% increase year-over-year, with strong growth in the data center (up 57%) and client and gaming segments (up 28%). Although there was a 3% sequential decline due to lower revenue in the embedded and data center segments, this was partially offset by growth in the client and gaming segments. The company's gross margin increased to 54%, and operating income reached $1.8 billion with a 24% operating margin. Diluted earnings per share rose 55% to $0.96. In the data center segment, revenue was driven by CPU server share gains and strong AMD Instinct GPU sales, despite a 5% sequential decline. The client and gaming segment saw strong demand for the Zen5 AMD Ryzen processors. Operating expenses increased as the company invested in market activities and R&D.

In the reported period, client revenue reached $2.3 billion, marking a 68% year-over-year increase, primarily due to higher average selling prices of high-end Ryzen processors. The client and gaming segment saw a 2% revenue rise sequentially, resulting in a $496 million operating income, up from $237 million a year earlier. The embedded segment revenue was $823 million, a 3% year-over-year decrease, with operating income down to $328 million. The company generated $939 million in operating cash and $727 million in free cash flow, returning $749 million to shareholders through stock repurchases. At quarter's end, the company held $7.3 billion in cash and equivalents and raised debt for an acquisition. The outlook for Q2 2025 anticipates approximately $7.4 billion in revenue, accounting for a new export license requirement impacting $700 million of revenue, still reflecting a 27% year-over-year growth at the midpoint.

In 2025, the company estimates a $1.5 billion revenue impact due to export license requirements. They anticipate double-digit growth in client and gaming revenues, flat performance in the embedded segment, and a decrease in data center revenue due to excluding MI308. Non-GAAP gross margin is expected to be 43%, impacted by $800 million inventory charges, but would be 54% without these charges. Operating expenses will be around $2.3 billion, including $50 million for the ZT System design. The ZT manufacturing business will report as discontinued from Q2, with $5 million in net interest and other expenses related to ZT transaction debt. The non-GAAP effective tax rate is projected at 13%, with a diluted share count of 1.64 billion, accounting for 9 million shares from the ZT transaction. Despite uncertainties, the company is optimistic about growth prospects, notably in AI and high-performance computing, and has started 2025 strongly with revenue growth, improved margins, and the acquisition of ZT Systems to bolster their data center offerings.

In the paragraph, Joshua Buchalter from TD Cowen asks about the factors driving better-than-expected results in both current performance and future guidance, particularly focusing on the client business, which grew 67% year-over-year. Lisa Su responds by highlighting strong performance across several sectors, including desktop performance in the client business, a successful Radeon launch in the gaming sector, and improvements in the data center business. She mentions that there has been a stronger product mix and increased average selling prices (ASPs), especially in the desktop gaming channel, which exceeded seasonal expectations. Looking ahead to Q2, they face a $700 million revenue headwind due to new export control limitations on the MI308.

The paragraph discusses the strong outlook of the business, highlighting growth and performance across various sectors. The desktop business and gaming sector are performing above typical seasonality, with console inventories depleted and starting a ramp-up. There's also growth in the data center segment despite challenges like China export controls. Jean Hu adds that strong Q1 performance in client revenue was due to increased average selling prices. Joshua Buchalter inquires about the performance of the Instinct family, and Lisa Su responds by indicating that the data center GPU performed slightly better than expected in Q1, expressing excitement for the upcoming MI350 launch.

In the article, it is mentioned that the MI350 launch is on track for mid-year, with strong customer interest and competitiveness due to its similarity to the MI300. There are important deals, including a collaboration with Oracle, expected to boost the AI business and lead to significant growth in the second half of the year, despite uncertainties like tariffs. In a subsequent discussion, Timothy Arcuri from UBS inquires about the growth of data center GPU sales, noting expectations of a significant increase from $600 million last March. Lisa Su addresses this by confirming that the data center GPU business performed well in the first quarter and most of the impact of an $800 million reduction from a ban will occur in the September quarter.

The paragraph is a discussion about AMD's current and future product lines and inventory management. It begins with Jean explaining the distribution of a $1.5 billion revenue focus over Q2, Q3, and a minimal Q4, primarily targeting non-China revenues focusing on the MI350 family. Timothy Arcuri inquires about the rise in inventory, which Jean attributes to preparation for a strong ramp in client and server demand, as well as future data center GPU needs. Harlan Sur from JPMorgan asks about the MI400 series, highlighting upcoming competitive advantages and challenges in deploying Rack Scale platforms. Lisa Su responds by expressing excitement over both the upcoming MI350 and the future MI400 series, emphasizing AMD's proactive engagement with customers on its roadmap.

In the paragraph, the speaker discusses the strategic acquisition of ZT Systems to address Rack Scale Architecture and how it's been a timely move. They highlight ongoing efforts with ZT and customer design teams to plan future Rack Scale Systems and capitalize on the enthusiasm for the MI400 series. Harlan Sur praises AMD's strong year-over-year growth in the EPYC Enterprise and on-premise segments, noting a significant increase in server market share. Lisa Su attributes this success to AMD's robust product offerings, specifically the fifth-gen EPYC and a broadened product portfolio for Turin, as well as enhanced go-to-market strategies, including increased headcount and direct engagement with end users.

The paragraph features a discussion about the growth and challenges faced by an enterprise, particularly in the data center and GPU business. Jean Hu addresses Aaron Rakers's question concerning the expected flat revenue for the data center business in the first half of the year, noting a significant financial impact from China. Jean explains that the $1.5 billion impact is mainly in Q2 and Q3, but they anticipate a substantial ramp-up with the launch of the MI355, expecting strong year-over-year growth. Aaron also inquires about gross margin expectations, questioning if it will return to over 54% in the year's second half. Jean confirms this expectation, considering the current quarter's guidance and mixed attributes.

In the paragraph, the company is discussing its gross margin and the factors affecting it, explaining that though the mix is less favorable due to growth in the client and gaming sectors, they have drivers that could raise the gross margin. Improvements are noted in the client business due to a richer product mix and in the data center with expanded market share. The MI308 data center GPU has a low gross margin, but overall, they expect slight improvement in gross margin in the second half of the year. The discussion transitions to a question from Thomas O'Malley regarding the company's system-based architectures and interconnect capabilities, and whether they need to develop these internally or look elsewhere. Lisa Su is expected to answer this query.

The paragraph discusses the company's strategic investments and partnerships in optimizing system-level architectures involving CPU, GPU, networking, and rack-scale architecture. The speaker emphasizes the importance of interoperability and flexibility in meeting customer needs, especially in networking. Thomas O'Malley inquires about the company's market share gains and growth expectations for the client business in 2025. Lisa Su responds, noting satisfaction with the company's recent performance in the client business, highlighting unit growth in desktops and increased revenue share in high-end notebooks, commercial sectors, and overall desktops. The focus is on gaining share in high-value segments.

The paragraph discusses the company's focus on aligning with customers, managing inventory levels, and assessing overall consumption and sell-through trends amid macroeconomic factors and tariffs. The speaker, Lisa Su, addresses questions from analyst Vivek Arya about the company's GPU sales. She confirms that while data center GPU sales were slightly down in Q1 compared to Q4, they anticipate strong double-digit growth in data center GPUs, despite challenges like the MI308 situation and China headwinds. The strong demand in Q1 was driven by the MI325, with significant adoption by a major foundational model company. Jean Hu adds that MI325 and MI300 contributed significantly to their revenue in Q1.

In the paragraph, Lisa Su discusses the impact of China export controls and AI diffusion rules on the total addressable market (TAM) for AI accelerators, valued at approximately $500 billion. She mentions that the restrictions on leading-edge GPUs in China were anticipated and included in their TAM calculations, so they don't significantly alter the overall market size. She emphasizes their active engagement with the government on AI diffusion rules and aims to maintain a balance that allows global usage of US technology. Su stresses the importance of the world adopting the US AI ecosystem, given their technological leadership. After her response, the discussion moves on to the next question from C J Muse regarding assumptions around the client market.

In the paragraph, Lisa Su discusses the performance and expectations for the company's client business. She highlights that the strong product portfolio, particularly desktop channel products, is driving positive results, which deviate from traditional seasonality patterns. Although the first half of the year has performed better than usual, the company expects the second half to be sub-seasonal. Lisa mentions a strong Q1 performance and ongoing positive trends into Q2, largely due to an upgrade cycle and demand for gaming CPUs and GPUs. The company is ramping up production to meet demand. C J Muse then asks about the company's strategy for the 400 series and rack-level solutions, specifically inquiring about partnership targets for next year.

The discussion centers around overcoming learning curve challenges for rack scale solutions with OEM partners to deliver progress by 2026. Lisa Su emphasizes the importance of starting early and collaborating closely with hyperscale and OEM partners to define solutions and address potential issues. The focus is on advancing the learning cycle, particularly concerning the MI400 series. In a separate exchange, Stacy Rasgon asks about the potential growth of the GPU business in light of challenges in China. Lisa Su acknowledges that the GPU business will not grow year-over-year in Q2 due to a $700 million impact but expects growth in Q3 and Q4 to achieve strong double-digit growth for the full year.

In the paragraph, an analyst, Stacy, asks about trends in Q1, noting that data center GPU revenue was down modestly as expected, but seems to imply a sequential decline of at least 20%, which also suggests server CPUs were up sequentially, surpassing seasonal expectations. Jean Hu responds, confirming a 5% decline in data center performance but indicates a slightly better situation for servers, which also declined sequentially. The conversation then shifts to Ross Seymore from Deutsche Bank, who inquires about the embedded sector's outlook. Jean Hu mentions seeing signs of gradual recovery, with improving order patterns and book-to-bill ratios, giving confidence in achieving significant sequential growth in the second half of the year.

The paragraph covers a discussion on improved demand trends in the aerospace, defense, and test measurement sectors, while industrial improvement is less visible due to existing inventory. There's an expectation of revenue growth year-over-year, especially in Q4. Ross Seymore inquires about OpEx guidance, particularly concerning the $50 million from ZT. Jean Hu confirms the incremental OpEx and mentions disciplined management of OpEx relative to revenue growth. The final question from Joe Moore discusses cloud customers' increasing inference costs and whether this affects AMD's business focus, with Lisa Su poised to respond.

The paragraph discusses the evolving landscape of AI models and inferencing, highlighting the importance of distributed inferencing and the strengths of AMD's MI300 and 353 series as inference solutions. AMD's advantages include memory bandwidth and capacity, and the company is focusing on optimizing for multiple models using their ROCm software suite due to the trend of using various models. Lisa Su comments on the competition with custom silicon (A6) in the AI space, arguing that while A6 has its place, GPUs have a larger potential market. She emphasizes the significance of offering competitive total cost of ownership (TCO) and providing choices for customers, especially as inference costs increase. Su suggests that it's not a matter of choosing solely between AMD and A6, but rather that both have roles in the AI ecosystem.

The paragraph indicates a discussion about being competitive in inferencing and improving in training solutions. It also notes the conclusion of a question-and-answer session and the teleconference. Participants are thanked and told they may now disconnect.

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