AMD: DeepSeek Is A Big Deal

Seeking Alpha01-22 14:37

SummaryAMD shares have fallen 23.37% since October, mainly due to concerns about competition with Nvidia's Blackwell chips, but AMD can still compete effectively.Breakthroughs from DeepSeek V3 model ...

Source Link

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Comments

We need your insight to fill this gap

Leave a comment

{"i18n":{"language":"en_US"},"isChannel":false,"data":{"share":"https://ttm.financial/m/news/1149826257?lang=en_US&edition=fundamental","thumbnail":"","is_english":true,"pubTime":"2025-01-22 14:37","share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","id":"1149826257","market":"us","top_or_hot":-1,"title":"AMD: DeepSeek Is A Big Deal","media":"Seeking Alpha","content":"<div>\n<p>SummaryAMD shares have fallen 23.37% since October, mainly due to concerns about competition with Nvidia's Blackwell chips, but AMD can still compete effectively.Breakthroughs from DeepSeek V3 model ...</p>\n\n<a href=\"https://seekingalpha.com/article/4750764-amd-stock-deepseek-is-a-big-deal\">Source Link</a>\n\n</div>\n","source":"lsy1728464409321","html":"<!DOCTYPE html>\n<html>\n<head>\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset=utf-8\" />\n<meta name=\"viewport\" content=\"width=device-width,initial-scale=1.0,minimum-scale=1.0,maximum-scale=1.0,user-scalable=no\"/>\n<meta name=\"format-detection\" content=\"telephone=no,email=no,address=no\" />\n<title>AMD: DeepSeek Is A Big Deal</title>\n<style type=\"text/css\">\na,abbr,acronym,address,applet,article,aside,audio,b,big,blockquote,body,canvas,caption,center,cite,code,dd,del,details,dfn,div,dl,dt,\nem,embed,fieldset,figcaption,figure,footer,form,h1,h2,h3,h4,h5,h6,header,hgroup,html,i,iframe,img,ins,kbd,label,legend,li,mark,menu,nav,\nobject,ol,output,p,pre,q,ruby,s,samp,section,small,span,strike,strong,sub,summary,sup,table,tbody,td,tfoot,th,thead,time,tr,tt,u,ul,var,video{ font:inherit;margin:0;padding:0;vertical-align:baseline;border:0 }\nbody{ font-size:16px; line-height:1.5; color:#999; background:transparent; }\n.wrapper{ overflow:hidden;word-break:break-all;padding:10px; }\nh1,h2{ font-weight:normal; line-height:1.35; margin-bottom:.6em; }\nh3,h4,h5,h6{ line-height:1.35; margin-bottom:1em; }\nh1{ font-size:24px; }\nh2{ font-size:20px; }\nh3{ font-size:18px; }\nh4{ font-size:16px; }\nh5{ font-size:14px; }\nh6{ font-size:12px; }\np,ul,ol,blockquote,dl,table{ margin:1.2em 0; }\nul,ol{ margin-left:2em; }\nul{ list-style:disc; }\nol{ list-style:decimal; }\nli,li p{ margin:10px 0;}\nimg{ max-width:100%;display:block;margin:0 auto 1em; }\nblockquote{ color:#B5B2B1; border-left:3px solid #aaa; padding:1em; }\nstrong,b{font-weight:bold;}\nem,i{font-style:italic;}\ntable{ width:100%;border-collapse:collapse;border-spacing:1px;margin:1em 0;font-size:.9em; }\nth,td{ padding:5px;text-align:left;border:1px solid #aaa; }\nth{ font-weight:bold;background:#5d5d5d; }\n.symbol-link{font-weight:bold;}\n/* header{ border-bottom:1px solid #494756; } */\n.title{ margin:0 0 8px;line-height:1.3;color:#ddd; }\n.meta {color:#5e5c6d;font-size:13px;margin:0 0 .5em; }\na{text-decoration:none; color:#2a4b87;}\n.meta .head { display: inline-block; overflow: hidden}\n.head .h-thumb { width: 30px; height: 30px; margin: 0; padding: 0; border-radius: 50%; float: left;}\n.head .h-content { margin: 0; padding: 0 0 0 9px; float: left;}\n.head .h-name {font-size: 13px; color: #eee; margin: 0;}\n.head .h-time {font-size: 11px; color: #7E829C; margin: 0;line-height: 11px;}\n.small {font-size: 12.5px; display: inline-block; transform: scale(0.9); -webkit-transform: scale(0.9); transform-origin: left; -webkit-transform-origin: left;}\n.smaller {font-size: 12.5px; display: inline-block; transform: scale(0.8); -webkit-transform: scale(0.8); transform-origin: left; -webkit-transform-origin: left;}\n.bt-text {font-size: 12px;margin: 1.5em 0 0 0}\n.bt-text p {margin: 0}\n</style>\n</head>\n<body>\n<div class=\"wrapper\">\n<header>\n<h2 class=\"title\">\nAMD: DeepSeek Is A Big Deal\n</h2>\n\n<h4 class=\"meta\">\n\n\n2025-01-22 14:37 GMT+8&nbsp;&nbsp;&nbsp;<a href=https://seekingalpha.com/article/4750764-amd-stock-deepseek-is-a-big-deal><strong>Seeking Alpha</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>SummaryAMD shares have fallen 23.37% since October, mainly due to concerns about competition with Nvidia's Blackwell chips, but AMD can still compete effectively.Breakthroughs from DeepSeek V3 model ...</p>\n\n<a href=\"https://seekingalpha.com/article/4750764-amd-stock-deepseek-is-a-big-deal\">Source Link</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","isBrief":false,"type":0,"news_type":1,"symbol":"AMD","symbol_name":"美国超微公司","start_time":0,"source_url":"https://seekingalpha.com/article/4750764-amd-stock-deepseek-is-a-big-deal","article_id":"1149826257","we_media_id":null,"thumbnails":[],"rights":null,"url":"https://stock-news.laohu8.com/highlight/detail?id=1149826257","pubTimestamp":1737527842,"columns":[],"sourceInfo":{"source_id":"lsy1728464409321","name":"Seeking Alpha"},"weMediaInfo":null,"summary":"SummaryAMD shares have fallen 23.37% since October, mainly due to concerns about competition with Nvidia's Blackwell chips, but AMD can still compete effectively.Breakthroughs from DeepSeek V3 model s","collect":0,"end_time":0,"defaultTopTitle":"seekingalpha.com","property":[],"viewcount":null,"language":"en","relate_stocks":{"AMD":"美国超微公司"},"translate_title":"AMD：DeepSeek是一件大事","themeId":"","isJumpTheme":false,"ttsUrl":null,"symbols_score_info":{"AMD":1.1},"content_text":"SummaryAMD shares have fallen 23.37% since October, mainly due to concerns about competition with Nvidia's Blackwell chips, but AMD can still compete effectively.Breakthroughs from DeepSeek V3 model significantly reduce AI training costs, positioning AMD GPUs as an even more cost-effective alternative to Nvidia's, enhancing AMD's market position.AMD's forward P/E ratio is undervalued, and aligning its PEG ratio with the sector median could lift its valuation by approximately 130.49%.Despite risks, AMD's competitive total cost of ownership and DeepSeek's advancements make AMD shares a strong buy for future growth in the AI GPU market.undefined undefinedInvestment ThesisAdvanced Micro Devices (NASDAQ:AMD) (NEOE:AMD:CA) shares are down -23.37% since the last time I wrote on the CPU and GPU maker in late October before their Q3 earnings. I think one of the main reasons shares have sold off since the fall has been the market’s concern that AMD is not catching up fast enough to Nvidia’s new Blackwell chips. I continue to believe Blackwell is a powerful product, but also something AMD can and will successfully compete against.In the meantime, even CEO Lisa Su admitted in October that despite their efforts to compete against NVIDIA's (NVDA) Blackwell chips through their MI350 series, growth opportunities in the future are still “lumpy.”Over the last few weeks, however, research out of China has changed the game. I think a new model, named DeepSeek V3, has the potential to totally change the training and inference game. AMD will be able to benefit.With OpenAI announcing their latest model, o3, in December, I think the industry is going to increasingly focus on the cost of compute. Some queries on o3 cost OpenAI north of $1000.In essence, breakthroughs by DeepSeek are going to allow AI labs to exponentially decrease the amount of compute needed to generate strong LLMs. With less cutting edge compute needed, they will be able to optimize for cost. I believe AMD’s GPUs are part of this answer.With this, I continue to be a strong buy.Why I'm Doing Follow-Up CoverageSince the last time I covered AMD, I believe we’ve seen an incredible development in the large language models due to DeepSeek figuring out how to more efficiently train models. DeepSeek’s development comes right after post-training inferencing for models such as OpenAI new models and the new Google’s Gemini model were announced.With this, I think the attention on GPUs is going to move to the cost per query. Models that require multiple passes to do multi-step reasoning inherently cost more per query than a model that just does one pass because they process more tokens or require more iterations, which increases the cost per query. With this, I think it’s going to be incredibly important to focus on cost components, and AMD has an incredible position with this. DeepSeek magnifies this position.Hyper-scalers, such as Google, Microsoft, and AWS, all operate with extreme sensitivity to costs because of their enormous scales. Gartner predicts that these hyper-scalers will be spending up to $500 billion on AI servers alone by 2028. As such, their focus will always be on how they can bring down operational costs, which include hardware expenses as well as energy consumption and processing efficiency.I’m doing a follow-up post to show how this new model training design in 2025 will be a game changer for LLM companies. I continue to believe that lower LLM training budgets benefits AMD specifically.DeepSeek V3 Deep DiveResearchers from Stanford University and Epoch AI published a study mid last year showing that the cost of training frontier AI models has grown by a factor of two to three times annually for the past eight years. They expect that the largest models will cost over $1 billion to train by 2027. This is proving to be a huge barrier to entry in the space and is making it so that after each successive model training iteration, only a really small group of companies can compete.Ben Cottier from Epoch AI warned in a piece in Time Magazine that only large companies will be able to compete in the industry because of the costs involved:The cost of the largest AI training runs is growing by a factor of two to three per year since 2016, and that puts billion-dollar price tags on the horizon by 2027, maybe sooner.In December, DeepSeek launched its latest model, the V3, which looks like it could be a key solution to this problem.What makes DeepSeek so important is its training cost relative to its performance. DeepSeek’s latest model benchmarks well compared to one of OpenAI’s strongest models (GPT 4o).DeepSeek V3 Performance (GitHub)Yet, the cost to train this model was only $5.6 million. This is far less than US-based models. This training architecture represents a breakthrough because it means that leading LLM labs can use GPUs other than Nvidia’s leading GPU (Blackwell) in order to build competitive models.This is where, in my opinion, AMD comes in. AMD’s latest GPU (the MI 325X) cost far less (roughly $15,000) compared to the cost of Nvidia’s latest generation of Blackwell chips (up to $70,000).Before, this mattered because it was a question of optimizing for price or model training capabilities. Now, with DeepSeek’s architecture, it doesn't matter.With this, AMD’s Instinct MI325X is said to perform around the same level as Nvidia’s H100s. Later generations of AMD chips (MI 350) are rumored to perform well compared to where Blackwell is performing.If we look at the company's core data center business, the company booked an impressive $3.5 billion in data center revenue in Q3, up 122% YoY. I continue to believe this is going to be an incredible growth driver even for AMD through the rest of the decade. Now, DeepSeek provides another key catalyst.ValuationAlthough AMD shares have fallen since the last time I wrote on the GPU company in October, I still think we need to look at their outlook: the company still has incredible growth ahead of them. AMD shares currently trade at roughly 24.35 times the forward earnings estimate for this year (2025), with their EPS projected to come in at $4.99/share.What’s key about this is that this forward P/E is on the lower end of AMD’s historical forward P/E range. The market is expecting earnings growth to be far slower for the GPU maker than it’s been for its larger rival, Nvidia, even though I believe they haven’t hit scale in the GPU market.AMD Historical PE (Macrotrends)This tells me that the market is not confident in AMD. The company’s forward PEG ratio tells me that as well. Their forward PEG ratio of 0.82 is less than half the sector median’s forward PEG ratio of 1.89.At the same time, the company has revised their AI GPU sales guidance upward multiple times in 2024, increasing projections from $2 billion to $5 billion. This demonstrates AMD’s accelerating position in the growing AI GPU market. Unfortunately, it’s the legacy segments of AMD’s business—namely, consumer CPUs and GPUs—that remain a drag on their valuation.Yet, I still believe this core AI GPU part of the business that will be extremely valuable over the next 5 to 10 years as more lightweight LLM training and inference clusters are set up.DeepSeek really changed the game for how GPU clusters can be set up. I think the industry can focus more on cost now. AMD should benefit from this.I think AMD’s PEG ratio should align with the sector median of 1.89. If we saw this happen (and their PEG ratio expanded from its current 0.82), this would mean their valuation could experience an uplift of approximately 130.49% from its current levels.RisksI think one of the biggest risks for AMD despite this DeepSeek news is that their GPUs may not be able to catch up to Nvidia’s latest chip lineup. This doesn’t mean that AMD will not be able to catch up to Blackwell. This just means they will not be able to match Nvidia in terms of GPU capabilities at any one time.CEO Su admitted in October that they have ramped up their development efforts, although this could strain their resources:We feel very good about the growth opportunities. I would say that it might be lumpy. In general, these are large customer acquisitions and it’s not always predictable exactly which quarters you would expect the significant build-out…we have successfully accelerated our product development pace to deliver an annual cadence of new Instinct products. Our next-gen MI350-series silicon is looking very good and is on track to launch in the second half of 2025 with the largest generational increase in AI performance we have ever delivered.To offset this risk, I think it's key to go back to the main conclusions of the DeepSeek announcement. What’s key about DeepSeek is that this new paper proved that you don’t need the state-of-the-art chips to produce state-of-the-art results.So even if AMD’s chips don’t match Nvidia's, that's okay. This new paper shows that the use cases for AMD’s chips have exponentially expanded. If we even look at the DeepSeek V3 model itself, what’s key about this model is that it was trained on Nvidia H800s, which are slower GPUs that the company had customized for Chinese customers to get around export restrictions.Nvidia’s H800 (what DeepSeek was trained on) is slower than its H100 GPU. AMD’s MI 300X GPU outperforms the H100 in key benchmarks. Therefore, DeepSeek’s paper helps show us that this new model training architecture can support state-of-the-art model training on AMD GPUs. This is a big deal.Bottom LineWhile AMD shares have trailed the market since I last wrote on the GPU maker, I still believe they’re set for strong growth, especially given their competitive total cost of ownership and now with DeepSeek showing LLM developers how to build state-of-the-art models for a fraction of the price. With their GPUs coming in at a lower cost than Nvidia's, DeepSeek means AMD can now offer comparable performance at much lower prices. This is huge for AI. This is big for AMD.With this, I continue to be bullish. I believe shares are a strong buy.","kind":"news","is_publish_news":true,"is_publish_highlight":true,"is_publish_live":false,"is_publish_wemedia":null,"editions":null,"column":"","sentiment":"0","news_tag":"","news_rank":0,"symbols":[],"gpt_button":0},"commentList":[],"isCommentEnd":true,"newsSizeData":{"likeSize":1,"commentSize":0,"repostSize":0,"favoriteSize":0,"likeStatus":false,"favoriteStatus":false},"APP":{"userAgent":"Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; ClaudeBot/1.0; +claudebot@anthropic.com)","isDev":false,"isTTM":true,"deviceId":"web-server-community-laohu8-v3","version":"4.30.2","shortVersion":"4.30.2","platform":"web","vendor":"web","appName":"ttm","isIOS":false,"isAndroid":false,"isTiger":false,"isTHS":false,"isWeiXin":false,"isWeiXinMini":false,"isWeiBo":false,"isQQ":false,"isBaiduSwan":false,"isBaiduBox":false,"isDingTalk":false,"isToutiao":false,"isOnePlus":false,"isHuaWei":false,"isXiaomi":false,"isXiaomiWebView":false,"isOppo":false,"isVivo":false,"isSamsung":false,"isMobile":false},"href":"/m/news/1149826257"}