$Microsoft(MSFT)$ Microsoft un

$Microsoft(MSFT)$ Microsoft unveiled Maia 300, saying Maia will become its primary inference system, with multi-gigawatt deployments starting in 2027.

Most of the upgrade is in memory. Compared with Maia 200, compute FLOPs rise 36%, HBM bandwidth rises 2.5x to 18 TB/s, and capacity rises 33% to 288 GB.

Microsoft expects 40% more tokens per megawatt than leading GPU-based systems, though it doesn't name the system it's comparing against. Full racks double to 96 accelerators, which helps get to the 7x rack throughput figure.

Production deployments start in 2027. Evercore ISI estimates Maia will still be only about 10% of Microsoft's AI chip spending by the first quarter of 2027 (via CNBC).

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"},"detailType":1,"isChannel":false,"data":{"magic":2,"id":617939252085376,"tweetId":"617939252085376","gmtCreate":1791823347000,"gmtModify":1791873202248,"author":{"id":10000000000010988,"idStr":"10000000000010988","authorId":10000000000010988,"authorIdStr":"10000000000010988","name":"Marian Horace","avatar":"https://static.tigerbbs.com/45889c069df88784e75b651f763a062a","vip":8,"userType":8,"introduction":"","boolIsFan":false,"boolIsHead":false,"crmLevel":1,"crmLevelSwitch":0,"individualDisplayBadges":[],"wearingBadges":[],"fanSize":44,"starInvestorFlag":false},"themes":[],"images":[],"coverImages":[],"title":"","html":"<html><head></head><body><a href=\"https://ttm.financial/S/MSFT\">$Microsoft(MSFT)$&nbsp;</a>Microsoft unveiled Maia 300, saying Maia will become its primary inference system, with multi-gigawatt deployments starting in 2027.\nMost of the upgrade is in memory. Compared with Maia 200, compute FLOPs rise 36%, HBM bandwidth rises 2.5x to 18 TB/s, and capacity rises 33% to 288 GB.\nMicrosoft expects 40% more tokens per megawatt than leading GPU-based systems, though it doesn't name the system it's comparing against. Full racks double to 96 accelerators, which helps get to the 7x rack throughput figure.\nProduction deployments start in 2027. Evercore ISI estimates Maia will still be only about 10% of Microsoft's AI chip spending by the first quarter of 2027 (via CNBC).</body></html>","htmlText":"<html><head></head><body><a href=\"https://ttm.financial/S/MSFT\">$Microsoft(MSFT)$&nbsp;</a>Microsoft unveiled Maia 300, saying Maia will become its primary inference system, with multi-gigawatt deployments starting in 2027.\nMost of the upgrade is in memory. Compared with Maia 200, compute FLOPs rise 36%, HBM bandwidth rises 2.5x to 18 TB/s, and capacity rises 33% to 288 GB.\nMicrosoft expects 40% more tokens per megawatt than leading GPU-based systems, though it doesn't name the system it's comparing against. Full racks double to 96 accelerators, which helps get to the 7x rack throughput figure.\nProduction deployments start in 2027. Evercore ISI estimates Maia will still be only about 10% of Microsoft's AI chip spending by the first quarter of 2027 (via CNBC).</body></html>","text":"$Microsoft(MSFT)$ Microsoft unveiled Maia 300, saying Maia will become its primary inference system, with multi-gigawatt deployments starting in 2027. Most of the upgrade is in memory. Compared with Maia 200, compute FLOPs rise 36%, HBM bandwidth rises 2.5x to 18 TB/s, and capacity rises 33% to 288 GB. Microsoft expects 40% more tokens per megawatt than leading GPU-based systems, though it doesn't name the system it's comparing against. Full racks double to 96 accelerators, which helps get to the 7x rack throughput figure. Production deployments start in 2027. Evercore ISI estimates Maia will still be only about 10% of Microsoft's AI chip spending by the first quarter of 2027 (via CNBC).","highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"favoriteSize":0,"link":"https://ttm.financial/post/617939252085376","repostId":0,"isVote":1,"tweetType":1,"viewCount":149,"commentLimit":10,"likeStatus":false,"favoriteStatus":false,"reportStatus":false,"symbols":["MSFT"],"verified":2,"subType":0,"readableState":1,"langContent":"EN","currentLanguage":"EN","warmUpFlag":false,"orderFlag":false,"shareable":true,"causeOfNotShareable":"","featuresForAnalytics":[],"commentAndTweetFlag":false,"andRepostAutoSelectedFlag":false,"upFlag":false,"length":586,"optionInvolvedFlag":false,"subscribersOnly":false,"subscribersOnlyAccessible":false,"xxTargetLangEnum":"ORIG"},"commentList":[],"isCommentEnd":true,"isTiger":false,"isWeiXinMini":false,"url":"/m/post/617939252085376"}