$AMD Q2 2024 AI-Generated Earnings Call Transcript Summary

AMD

Jul 31, 2024

The operator welcomes participants to the AMD Second Quarter 2024 Conference Call and introduces Mitch Haws, Vice President Investor Relations. Mitch reminds participants that the call is being recorded and directs them to review the earnings press release and accompanying slides on the Investor Relations page of amd.com. He also mentions that the call will primarily refer to non-GAAP financial measures and that the full reconciliations can be found in the press release and slides. Dr. Lisa Su, Chair and CEO, and Jean Hu, Executive Vice President, CFO, and Treasurer, will be speaking on the call. Mitch notes that the discussion will contain forward-looking statements and refers participants to the cautionary statement in the press release. He then hands the call over to Lisa.

In the second quarter, AMD saw strong financial results with revenue above expectations and profits increasing. This was driven by high sales of their Instinct, Ryzen, and EPYC processors, as well as increased demand for general purpose computing in both client and server businesses. Data center segment revenue reached a record high of $2.8 billion, with a significant increase in sales of EPYC CPUs and adoption by hyperscalers. AMD-powered cloud instances have also increased by 34% and are being chosen by companies like Netflix and Uber for their critical workloads.

In the enterprise sector, sales increased significantly and multiple large deals were closed with various industries. The next-generation Turin family, featuring the new Zen 5 core, is expected to offer improved performance and efficiency. Production has started for Turin and it is expected to be widely available later this year. In the data center AI business, there was record revenue for the third consecutive quarter, with Microsoft expanding their use of MI300X Accelerators for various services. Microsoft also became the first large hyperscaler to offer public MI300X instances.

The new Azure VMs use MI300X and ROCm software to provide top performance for inferencing, with Hugging Face being an early adopter. MI300 solutions are in high demand and are being made available through partnerships with companies like Dell, HPE, Lenovo, and Supermicro. AMD has also made progress in enhancing their AI software stack and collaborating with the open source community to support the latest AI algorithms and models. MI300 is the first GPU to support Stable Diffusion's SD 3.0 image generation LLM and multiple partners have announced support for the latest Llama 3.1 models using ROCm and MI300X.

AMD's Llama 3.1 is compatible with MI300 accelerators and can run the FP16 version of the 405B model on a single server, providing cost benefits. The recent acquisition of Silo AI expands their capabilities in serving large enterprise customers and developing language models. AMD is also investing in other AI companies and has a roadmap for future AI accelerators, with the MI325X set to launch this year, followed by the MI350 series in 2025 and the MI400 series in 2026.

AMD's AI solutions work has received positive feedback from customers, with several major companies joining to establish a standardized approach for building high-performance data centers and AI solutions. The company now expects data center GPU revenue to exceed $4.5 billion in 2024. In the client segment, revenue increased by 49% due to strong demand for previous generation Ryzen processors and initial shipments of the next generation Zen 5 processors. These processors offer significant improvements in instructions per clock and will be available for both desktop and notebook applications. The first Ryzen AI 300 series notebooks have received strong reviews and over 100 more are set to launch from various manufacturers.

In the gaming segment, revenue declined due to a decline in semi-custom SoC sales, but there were improvements in gaming graphics revenue. The embedded segment also saw a decrease in revenue, but there are signs of improvement and strong design wins for the future. The company also announced the retirement of Victor Peng, who has made significant contributions to the company.

The speaker thanks Victor for his contributions to AMD and wishes him well in retirement. They also discuss the company's strong second quarter results and their future growth opportunities, particularly in the data center and AI markets. They are increasing investments in hardware, software, and solutions to meet this demand. The speaker then turns the call over to Jean to provide more details on the financial results and outlook for the third quarter. The data center segment saw significant growth compared to the previous year.

In the second quarter of 2024, AMD saw significant growth in client segment revenue and expanded gross margin by 340 basis points. Data center revenue was up 115% year-over-year and accounted for nearly 50% of total revenue, driven by strong sales of AMD Instinct GPUs and EPYC servers. The client segment also saw a 49% increase in revenue, driven by sales of AMD Ryzen processors. Operating income for both segments saw significant increases compared to the previous year.

In the second quarter, the gaming segment revenue decreased by 59% due to inventory digestion and lower demand. The embedded segment revenue also decreased by 41% due to inventory normalization. Cash flow was positive, with $593 million generated from operations and $439 million in free cash flow. Inventory increased by $339 million to support the ramp of a data center GPU product. In the third quarter, revenue is expected to grow by 15% sequentially and 16% year-over-year, driven by strong growth in the data center and client segments. The non-GAAP gross margin is expected to be approximately 53.5%.

The company expects non-GAAP operating expenses to be around $1.9 billion, a non-GAAP effective tax rate of 13%, and a diluted share count of approximately 1.64 billion shares. They also plan to close the acquisition of Silo AI for $665 million in cash during the third quarter. The company had a successful quarter, with record MI300 revenue exceeding $1 billion and strong performance from their next-gen Ryzen and EPYC products. They anticipate continued growth and opportunities in the future, while maintaining financial discipline and operational excellence. The audience was invited to ask questions during the Q&A session. One question asked about the outlook for MI300 for the rest of the year and potential for rapid growth next year. The company is pleased with the progress of MI300 and sees potential for growth in the future.

The company is pleased with the progress of their products in the market and expects revenue to continue to ramp up in the second half of the year. They also have a strong roadmap for future product releases and are optimistic about the AI market's need for more compute. They are seeing good traction with their hardware and software solutions and have a growing pipeline of customers.

Lisa Su, CEO of AMD, is pleased with the progress of their EPYC server CPU. They have seen broad adoption across cloud and are focused on enterprise and third-party cloud instances. They have also seen traction in enterprise and third-party cloud adoption. Going into the second half of the year, they feel good about the market improving and the launch of Turin. They expect to see revenue from Turin in the second half of the year and are confident in their ability to continue gaining share in the server market. The next question comes from Timothy Arcuri with UBS.

Timothy Arcuri asks Lisa Su about the Data Center GPU roadmap. He wants to know if the greater than $4.5 billion revenue includes any from the upcoming 325 launch, and if she can provide more details about the 350 series. Lisa Su confirms that there will be a small contribution from 325 in the fourth quarter, with a ramp-up in the first half of next year. She also mentions that the 350 series will have multiple SKUs, with a range of options from air-cooled to liquid-cooled. The series is designed to fit into the same infrastructure as the MI300 series, allowing for a fast ramp-up for customers who have already invested in those products. The next question comes from Ross Seymore about the company's financials.

Ross Seymore asks Lisa Su about AMD's position in the AI PC market and any potential competition from ARM-based systems. Lisa Su responds that AMD is pleased with their Client business results and has a strong roadmap for launching Zen 5 based products in the middle of the year. She also mentions positive feedback on their products and expects above-typical seasonality in the second half of the year. She acknowledges competition from ARM-based systems, but believes that AMD's products are well positioned in the PC market.

Lisa Su, CEO of AMD, discusses the parallels between the company's current Instinct portfolio and their EPYC product launch from five to six years ago. She acknowledges that there was initially both positive and negative reactions to EPYC, but it served as a toehold for long-term customer engagement. Su believes that the same applies to the Instinct portfolio, with a focus on gaining more opportunities and larger deployments over time. AMD has already achieved key milestones this year, such as getting their hardware in volume with multiple hyperscalers and large Tier 2 customers, and their software has also matured substantially.

The speaker discusses the progress of ROCm and the company's continued investment in networking and long-term conversations. They acknowledge that there is still more work to be done, but are optimistic about the future. The speaker also addresses the ongoing debate about AI monetization and how AMD plans to compete with other suppliers in the industry. They mention that the industry as a whole recognizes the importance of investing in AI technology.

Lisa Su believes that the investment cycle for AI will remain strong and there will be multiple solutions for the market, including GPUs, custom chips, and ASICs. AMD has a strong roadmap and partnerships with key AI market makers, and deep engagement in hardware and software co-optimization. For large language models, GPUs are still the preferred architecture. MI300 has seen early deployments in inference, but training is also picking up, with expectations for inference to be larger than training in the market.

AMD expects both inference and training to be growth opportunities for them. They have made great progress in ramping up supply for the MI300 and exceeded $1 billion in the second quarter. However, the overall supply chain remains tight and will continue to be tight through 2025. AMD is working to build additional capacity and capability to meet demand. They expect to continue ramping up supply in the second half of the year.

AMD's Q3 guidance indicates close to $1 billion of revenue growth across Data Center, Client, and Embedded businesses. The company expects the majority of this growth to come from Data Center, followed by Client and Embedded.

Lisa Su discusses the current state of the gaming business, which has seen a double-digit decline. She breaks down the data center segment, which is the largest piece, followed by client and then single-digit sequential growth in the embedded side. When asked about the split within the data center segment, Su mentions that both Instinct GPUs and server growth will contribute to the overall growth. The next question is about the gap between AMD's Instinct products and its competitors in terms of rack level performance. Su talks about the increasing complexity of systems and the need for solutions like Infinity Fabric and UALink. She is pleased with the partners involved in UALink and believes it is an important capability.

Lisa Su, CEO of AMD, discusses the company's plans for the future and how they plan to help customers get to market faster. They have all the necessary components under their umbrella, including Infinity Fabric and networking capabilities from the acquisition of Pensando. They plan to invest more in this area and have already seen success with their MI300 supercomputing product, which has been mainly used by government and AI customers. The biggest hurdle for ramping up the next wave of customers is not memory issues, despite the noise in the system, but rather investing in software and addressing challenges in the hardware market.

The speaker expresses pride in the team's rapid product ramp to $1 billion in the second quarter and ongoing success in the third and fourth quarters. They mention successful qualification of multiple suppliers for HBM3 memory and plans to qualify HBM3E for future products. The speaker also highlights the improved performance of ROCm and the company's efforts to expand its customer base through acquisitions and organic hiring. They emphasize the potential for continued growth and the need to build scale to seize this opportunity.

During a recent conference call, Lisa Su, CEO of AMD, discussed the company's performance with analysts. Blayne Curtis asked about the upcoming MI300 and Lisa thanked him before moving on to a question from Tom O'Malley about the Client division. Tom asked about the above-seasonal trends in the second half of the year, specifically the launch of Zen 5 desktops and notebooks and the potential impact of AI PC. Lisa confirmed that these were the primary reasons for the expected growth. The final question came from Chris Danely about gross margin, and Jean Hu responded that they had made progress in expanding it from 50% to 53.5% for Q3.

The primary driver for the company's growth is the Data Center business, which has seen a significant increase in revenue. This has contributed to gross margin expansion and is expected to continue to do so in the second half of the year. Other factors, such as the PC business and Embedded business, will also play a role in the company's gross margin, but the overall trend is expected to continue. The MI300 is expected to have a positive impact on the company's average gross margin in the long-term. Overall, the Data Center business is expected to continue driving gross margin expansion.

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

More Earnings