[Earnings Summary] AMD, Strong Performance in AI Semiconductors! Nvidia's Lead Still Far Away - Accelerated Pace Expected for 2025
$Advanced Micro Devices (AMD.US)$ Reed's Inc. announced earnings after the market closed on October 29th. The stock price fell 7.6% in after-hours trading.
◇The performance for the July-September quarter was driven by strong demand for AI semiconductors.exceeded market expectations, but 10-12 month revenue outlook is below market expectations.。
◇The company AI semiconductor revenue outlook for this yearfrom the original $4.5 billion to $5 billionupward revisionHowever, the extent of this upward revision fell short of offsetting concerns about the overall momentum of the company.and did not fully offset concerns about the slowdown in overall momentum $NVIDIA (NVDA.US)$it was undeniable that it paled in comparison to the momentum of。
◇The company AI semiconductor revenue outlook for this yearfrom the original $4.5 billion to $5 billionupward revisionHowever, the extent of this upward revision fell short of offsetting concerns about the overall momentum of the company.and did not fully offset concerns about the slowdown in overall momentum $NVIDIA (NVDA.US)$it was undeniable that it paled in comparison to the momentum of。
Earnings Highlights
● 7-9月期実績:予想上振れ
◇調整後EPS(1株当たり利益):0.92ドル、市場予想は0.92ドル
◇売上高:6.82 billionドル、前年同期比で18%増、市場予想は6.71 billionドル
● 7-9月期実績:予想上振れ
◇調整後EPS(1株当たり利益):0.92ドル、市場予想は0.92ドル
◇売上高:6.82 billionドル、前年同期比で18%増、市場予想は6.71 billionドル
・Increased revenue in the datacenter and client segments.However,there was a decrease in revenue in the gaming and embedded segments.partiallyoffseteach other.
・The datacenter division showed a strong performance with a 122% year-on-year increase in revenue. The client division also performed well with a 29% increase in revenue, but the gaming division saw a 69% decrease in revenue, and the embedded division experienced a 25% decrease.
・The datacenter division showed a strong performance with a 122% year-on-year increase in revenue. The client division also performed well with a 29% increase in revenue, but the gaming division saw a 69% decrease in revenue, and the embedded division experienced a 25% decrease.
Outlook for October-December period: Below financial estimates.
◇revenue: $7.5 billion ± $0.3 billion, slightly below the market expectation of $7.55 billion in the middle. The company expects the strong performance of the datacenter and client divisions to continue. For the gaming and embedded divisions, the company indicated that there will be more conservative growth compared to the previous quarter.
◇Adjusted gross profit margin: Approximately 54%, market expectations are 54.2%
◇revenue: $7.5 billion ± $0.3 billion, slightly below the market expectation of $7.55 billion in the middle. The company expects the strong performance of the datacenter and client divisions to continue. For the gaming and embedded divisions, the company indicated that there will be more conservative growth compared to the previous quarter.
◇Adjusted gross profit margin: Approximately 54%, market expectations are 54.2%
Revenue forecast for AI semiconductors in 2024.
◇The company has revised its AI semiconductor revenue outlook for this year from the previous $4.5 billion to $5 billion.Upward revisionHowever, the $0.5 billion increase in outlook is It did not offset the concern of slowing down the momentum of the overall business.。
◇Of the 5 billion dollars, the initial traction is mainly through inference.Primarily based on inference.However, there are several adoptions of training, and it is expected to continue growing over the next few quarters.One year later, I think a fairly balanced portfolio between training and inference will be completed (Lisa Su, CEO).
◇The company has revised its AI semiconductor revenue outlook for this year from the previous $4.5 billion to $5 billion.Upward revisionHowever, the $0.5 billion increase in outlook is It did not offset the concern of slowing down the momentum of the overall business.。
◇Of the 5 billion dollars, the initial traction is mainly through inference.Primarily based on inference.However, there are several adoptions of training, and it is expected to continue growing over the next few quarters.One year later, I think a fairly balanced portfolio between training and inference will be completed (Lisa Su, CEO).
(Nvidia, a competitor, gained an almost monopolistic position in the AI semiconductor market because its initial products were optimal for training. When launching AI datacenters, the demand for training was overwhelming. In subsequent operations, the demand for inference became significant. Nvidia indicated during its early 2024 earnings call that the ratio of inference had reached 40%. They demonstrated that their products are also being utilized for inference, dispelling some experts' concerns. Furthermore, from AMD's recent earnings reports, it is suggested that AMD has begun to capture demand in both training and inference. While not as dominant as Nvidia in the expanding market share, AMD has also shown that it is gaining a certain share.)
The main datacenter division is performing well.
● Entire datacenter division
◇Datacenter division revenue increased by 122% year-on-year in the July-September quarter, reaching a record high of 3.5 billion dollars. This exceeded the market estimate of 3.46 billion dollars.
◇It accounts for 52% of total revenue and for the first time on a quarterly basis, it represents more than half.
◇The strong performance of the datacenter division is mainly due to the significant increase in shipments of AI semiconductors, such as "Instinct GPU", and the increase in CPU revenue.
● Entire datacenter division
◇Datacenter division revenue increased by 122% year-on-year in the July-September quarter, reaching a record high of 3.5 billion dollars. This exceeded the market estimate of 3.46 billion dollars.
◇It accounts for 52% of total revenue and for the first time on a quarterly basis, it represents more than half.
◇The strong performance of the datacenter division is mainly due to the significant increase in shipments of AI semiconductors, such as "Instinct GPU", and the increase in CPU revenue.
● AI datacenter business(Mainly competing with NVIDIA)
◇Revenue from datacenter GPUs has rapidly expanded due to the increased adoption of MI300X (NVIDIA's competitor to H100) by cloud and OEM, bringing GPU business closer in scale to CPU business.
◇ $Microsoft (MSFT.US)$ and $Meta Platforms (META.US)$ Expanded the use of MI300X to strengthen internal workloads.
◇Revenue from datacenter GPUs has rapidly expanded due to the increased adoption of MI300X (NVIDIA's competitor to H100) by cloud and OEM, bringing GPU business closer in scale to CPU business.
◇ $Microsoft (MSFT.US)$ and $Meta Platforms (META.US)$ Expanded the use of MI300X to strengthen internal workloads.
(Microsoft and Meta have purchased a large quantity of NVIDIA products while also using AMD products. In the AI semiconductor industry, demand still exceeds supply.)
● Server CPU business(Mainly competing with Intel)
◇Enterprise orders are accelerating, with cloud providers expanding the use of EPYC CPUs across their entire infrastructure.
◇The company views that they were able to gain server CPU market share this quarter due to the initial introduction of the 5th generation EPYC processor.
◇Enterprise orders are accelerating, with cloud providers expanding the use of EPYC CPUs across their entire infrastructure.
◇The company views that they were able to gain server CPU market share this quarter due to the initial introduction of the 5th generation EPYC processor.
Roadmap for AI semiconductors
◇The next-generation MI325X GPU was released earlier this month. This has expanded the advantage in memory capacity and bandwidth, achieving up to 20% higher inference performance and competitive training performance compared to the H200 (NVIDIA product). Customer and partner interest in the MI325X is increasing.
◇Production and shipment of MI325X is scheduled to begin this quarter, starting from January to March 2025. $Dell Technologies (DELL.US)$ Ya $Hewlett Packard Enterprise (HPE.US)$ 、 $LENOVO GROUP LTD (LNVGY.US)$ 、 $Super Micro Computer (SMCI.US)$A wide range of systems will become available from various sources, etc.
◇In the long term, accelerate the product development pace and provide new Instinct products every year. The next-generation MI350 series is performing very well and is progressing smoothly towards its release in the second half of 2025.
◇The next-generation MI325X GPU was released earlier this month. This has expanded the advantage in memory capacity and bandwidth, achieving up to 20% higher inference performance and competitive training performance compared to the H200 (NVIDIA product). Customer and partner interest in the MI325X is increasing.
◇Production and shipment of MI325X is scheduled to begin this quarter, starting from January to March 2025. $Dell Technologies (DELL.US)$ Ya $Hewlett Packard Enterprise (HPE.US)$ 、 $LENOVO GROUP LTD (LNVGY.US)$ 、 $Super Micro Computer (SMCI.US)$A wide range of systems will become available from various sources, etc.
◇In the long term, accelerate the product development pace and provide new Instinct products every year. The next-generation MI350 series is performing very well and is progressing smoothly towards its release in the second half of 2025.
Key comments from CEO Lisa Su at the earnings call
◇ Market size of AI semiconductors
The rapid introduction of AI has led to a dramatic expansion, enabling new experiences where high-performance computing becomes even more essential in everyday life. In datacenters alone, the TAM (total addressable market) of AI semiconductors is expected to grow by over 60% annually, reaching $500 billion by 2028. This is equivalent to the annual revenue of the entire semiconductor industry in 2023.
◇ Market size of AI semiconductors
The rapid introduction of AI has led to a dramatic expansion, enabling new experiences where high-performance computing becomes even more essential in everyday life. In datacenters alone, the TAM (total addressable market) of AI semiconductors is expected to grow by over 60% annually, reaching $500 billion by 2028. This is equivalent to the annual revenue of the entire semiconductor industry in 2023.
◇ Competition with Nvidia
・At the time of release, MI300 was lagging behind H100. H100 had been in the market for a longer period. With the accelerated roadmap, most of that gap was actually filled. MI325 competes highly with H200 (Nvidia's next-generation product after H100), and MI350 competes strongly with Blackwell (Nvidia's latest AI semiconductor).
・On a global scale, the market continues to be constrained (demand exceeding supply), especially in new product generations where it takes a long time to ramp up actual production workloads after shipping the initial samples. I believe there are many opportunities across the entire set of AI workloads as our products are easily deployable.
・(When asked about the points where they are lagging behind compared to Nvidia) If I were to say, I see it as an opportunity. I think next year will be a year where both the customer base and workloads expand. And when we start on the MI400 series, I think it will be an exceptional product.
・At the time of release, MI300 was lagging behind H100. H100 had been in the market for a longer period. With the accelerated roadmap, most of that gap was actually filled. MI325 competes highly with H200 (Nvidia's next-generation product after H100), and MI350 competes strongly with Blackwell (Nvidia's latest AI semiconductor).
・On a global scale, the market continues to be constrained (demand exceeding supply), especially in new product generations where it takes a long time to ramp up actual production workloads after shipping the initial samples. I believe there are many opportunities across the entire set of AI workloads as our products are easily deployable.
・(When asked about the points where they are lagging behind compared to Nvidia) If I were to say, I see it as an opportunity. I think next year will be a year where both the customer base and workloads expand. And when we start on the MI400 series, I think it will be an exceptional product.
◇ AI semiconductor supply capability
Clearly, the supply environment is tough, but we have achieved significant results in securing production capacity across our entire supply chain. Over the next few quarters towards 2025, we expect the environment to remain challenging, but we also plan for substantial growth towards 2025. Therefore, we are satisfied with the overall capacity of our supply chain.
Clearly, the supply environment is tough, but we have achieved significant results in securing production capacity across our entire supply chain. Over the next few quarters towards 2025, we expect the environment to remain challenging, but we also plan for substantial growth towards 2025. Therefore, we are satisfied with the overall capacity of our supply chain.
◇ AI computer
The AI computer cycle is still in its early stages, but there is good momentum. In the client segment, we see AI computers as an opportunity for us. We believe the client segment is underestimated, and there is an opportunity for growth in both consumers and businesses.
The AI computer cycle is still in its early stages, but there is good momentum. In the client segment, we see AI computers as an opportunity for us. We believe the client segment is underestimated, and there is an opportunity for growth in both consumers and businesses.
◇ Outlook for 2025.
Roughly speaking about 2025, in terms of communication with customers, we are very satisfied with the market. The product portfolio is expected to strengthen the annual cycle with the release of MI325 in the latter part of this quarter, followed by MI355 in the latter part of next year.
Roughly speaking about 2025, in terms of communication with customers, we are very satisfied with the market. The product portfolio is expected to strengthen the annual cycle with the release of MI325 in the latter part of this quarter, followed by MI355 in the latter part of next year.
Target stock price by analysts.
After the earnings reports, several analysts have lowered their target stock prices. As of October 30, the average target stock price is $184.23.
After the earnings reports, several analysts have lowered their target stock prices. As of October 30, the average target stock price is $184.23.
Overall, Benefiting from the AI boom, AMD is also expected to gain a certain market share in the AI semiconductor market.Share.It is expected to be able to acquire. However, the momentum from this earnings report indicates that it is not as strong as the previous quarter.The short-term pace of business expansion.fell below analysts' expectations. CEO Lisa Su aims to strengthen the product cycle to improve.At the same time, she expressed confidence in Blackwell's competitors, the MI350 and subsequent MI400 series.However, given NVIDIA's strong momentum as a competitor, CEO Su will need to demonstrate the future performance.With the performance ahead,toprove thenecessity.ValuationWith a 12-month forward PE (price-to-earnings ratio) of 32.8 times, which is below the 5-year average of 35.1 times, it seems to be somewhat undervalued.
Created on October 30, 24 by Market Analyst Julie
Source: Company documents and Bloommberg, prepared by moomoo Securities
Source: Company documents and Bloommberg, prepared by moomoo Securities
Disclaimer: Moomoo Technologies Inc. is providing this content for information and educational use only.
Read more
Comment
Sign in to post a comment