The next AI bottleneck may not

The next AI bottleneck may not be compute—it may be connectivity.

ECOC 2026 is showing why: Coherent is demonstrating 3.2T pluggable optics and a 6.4T NPO engine, while Marvell is pushing 400G/lane technology toward 3.2T.

The deeper story is power per bit. As electrical links become harder to scale, optics are moving closer to the chip through NPO/CPO.

For investors, this expands the AI infrastructure map beyond GPUs into DSPs, lasers, silicon photonics, optical modules and packaging.

3.2T and 6.4T are still largely technology demonstrations—not mass deployment. But the direction is clear: AI may increasingly be limited by how fast machines can talk, not how fast they can compute.

The next AI arms race could be fought in photons, not transistors.

@Tiger_comments [思考]

@Tiger_comments：AI’s Next Arms Race Isn’t Just in GPUs — It’s in Optical Interconnects

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":610164137484568,"tweetId":"610164137484568","gmtCreate":1789992533052,"gmtModify":1789992538306,"author":{"id":4194184698932812,"idStr":"4194184698932812","authorId":4194184698932812,"authorIdStr":"4194184698932812","name":"苏36","avatar":"https://community-static.tradeup.com/news/default-avatar.jpg","vip":1,"userType":1,"introduction":"","boolIsFan":false,"boolIsHead":false,"crmLevel":12,"crmLevelSwitch":0,"currentWearingBadge":{"badgeId":"9427a906234c4b18a82ec3d7730b7fa8-1","templateUuid":"9427a906234c4b18a82ec3d7730b7fa8","name":"Star Contributor","description":"Published at least 3 PICK posts (or 10 IDEA posts) and, within the past 30 days, posted at least 1 PICK post (or 4 IDEA posts) and 1 comment","bigImgUrl":"https://community-static.tradeup.com/news/1866dcf97a73be1c330f85862546aedc","smallImgUrl":"https://community-static.tradeup.com/news/4f5c5fa8e2c7683bb5a7fce8753ee456","redirectLinkEnabled":1,"redirectLink":"https://ttm.financial/activity/market/2023/star-contributors/","hasAllocated":1,"isWearing":1,"stampPosition":0,"hasStamp":0,"allocationCount":1,"allocatedDate":"2026.09.17","individualDisplayEnabled":1,"backgroundColor":{"dark":"#2f2a1f","tint":"#fcf6e5"},"fontColor":{"dark":"#dea95e","tint":"#885600"}},"individualDisplayBadges":[{"badgeId":"9427a906234c4b18a82ec3d7730b7fa8-1","templateUuid":"9427a906234c4b18a82ec3d7730b7fa8","name":"Star Contributor","description":"Published at least 3 PICK posts (or 10 IDEA posts) and, within the past 30 days, posted at least 1 PICK post (or 4 IDEA posts) and 1 comment","bigImgUrl":"https://community-static.tradeup.com/news/1866dcf97a73be1c330f85862546aedc","smallImgUrl":"https://community-static.tradeup.com/news/4f5c5fa8e2c7683bb5a7fce8753ee456","redirectLinkEnabled":1,"redirectLink":"https://ttm.financial/activity/market/2023/star-contributors/","hasAllocated":1,"isWearing":1,"stampPosition":0,"hasStamp":0,"allocationCount":1,"allocatedDate":"2026.09.28","individualDisplayEnabled":1,"backgroundColor":{"dark":"#2f2a1f","tint":"#fcf6e5"},"fontColor":{"dark":"#dea95e","tint":"#885600"}}],"wearingBadges":[{"categoryType":2007,"badgeId":"9427a906234c4b18a82ec3d7730b7fa8-1","name":"Star Contributor","description":"Published at least 3 PICK posts (or 10 IDEA posts) and, within the past 30 days, posted at least 1 PICK post (or 4 IDEA posts) and 1 comment","smallImgUrl":"https://community-static.tradeup.com/news/4f5c5fa8e2c7683bb5a7fce8753ee456","bigImgUrl":"https://community-static.tradeup.com/news/1866dcf97a73be1c330f85862546aedc","isScarce":1,"plateColors":{"bgColor":{"dark":"#2f2a1f","tint":"#fcf6e5"},"fontColor":{"dark":"#dea95e","tint":"#885600"}},"effectEnabled":0,"redirectLinkEnabled":1,"redirectLinkType":1,"redirectLink":"https://ttm.financial/activity/market/2023/star-contributors/","redirectLinkValidityFrom":1679961600000,"redirectLinkValidityTo":1996963200000,"validityToTimestamp":1793118002902,"sortingWeight":0,"wearable":1}],"fanSize":1029,"starInvestorFlag":false},"themes":[],"images":[],"coverImages":[],"title":"","html":"<html><head></head><body>The next AI bottleneck may not be compute—it may be connectivity.\n\nECOC 2026 is showing why: Coherent is demonstrating 3.2T pluggable optics and a 6.4T NPO engine, while Marvell is pushing 400G/lane technology toward 3.2T.\nThe deeper story is power per bit. As electrical links become harder to scale, optics are moving closer to the chip through NPO/CPO.\nFor investors, this expands the AI infrastructure map beyond GPUs into DSPs, lasers, silicon photonics, optical modules and packaging.\n3.2T and 6.4T are still largely technology demonstrations—not mass deployment. But the direction is clear: AI may increasingly be limited by how fast machines can talk, not how fast they can compute.\nThe next AI arms race could be fought in photons, not transistors.\n<a href=\"https://ttm.financial/U/3501196737273098\"> @Tiger_comments </a> [思考]</body></html>","htmlText":"<html><head></head><body>The next AI bottleneck may not be compute—it may be connectivity.\n\nECOC 2026 is showing why: Coherent is demonstrating 3.2T pluggable optics and a 6.4T NPO engine, while Marvell is pushing 400G/lane technology toward 3.2T.\nThe deeper story is power per bit. As electrical links become harder to scale, optics are moving closer to the chip through NPO/CPO.\nFor investors, this expands the AI infrastructure map beyond GPUs into DSPs, lasers, silicon photonics, optical modules and packaging.\n3.2T and 6.4T are still largely technology demonstrations—not mass deployment. But the direction is clear: AI may increasingly be limited by how fast machines can talk, not how fast they can compute.\nThe next AI arms race could be fought in photons, not transistors.\n<a href=\"https://ttm.financial/U/3501196737273098\"> @Tiger_comments </a> [思考]</body></html>","text":"The next AI bottleneck may not be compute—it may be connectivity. ECOC 2026 is showing why: Coherent is demonstrating 3.2T pluggable optics and a 6.4T NPO engine, while Marvell is pushing 400G/lane technology toward 3.2T. The deeper story is power per bit. As electrical links become harder to scale, optics are moving closer to the chip through NPO/CPO. For investors, this expands the AI infrastructure map beyond GPUs into DSPs, lasers, silicon photonics, optical modules and packaging. 3.2T and 6.4T are still largely technology demonstrations—not mass deployment. But the direction is clear: AI may increasingly be limited by how fast machines can talk, not how fast they can compute. The next AI arms race could be fought in photons, not transistors. @Tiger_comments [思考]","highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"favoriteSize":0,"link":"https://ttm.financial/post/610164137484568","repostId":610131541590104,"repostType":1,"repost":{"magic":2,"id":610131541590104,"tweetId":"610131541590104","gmtCreate":1789988593563,"gmtModify":1789988675554,"author":{"id":3501196737273098,"idStr":"3501196737273098","authorId":3501196737273098,"authorIdStr":"3501196737273098","name":"Tiger_comments","avatar":"https://community-static.tradeup.com/news/227887b200e9925968650d5db4a8bfb3","vip":4,"userType":4,"introduction":"Tiger Official Account","boolIsFan":false,"boolIsHead":false,"hat":"https://static.tigerbbs.com/6444b4970948c1077534c593eccd6d16","crmLevel":1,"crmLevelSwitch":0,"individualDisplayBadges":[],"wearingBadges":[],"fanSize":8786,"starInvestorFlag":false},"themes":[{"themeId":"228f7dbe92224ab7969df668216f4d7f","categoryId":"7a15628ce2174325bddc6229505e0d70","name":"Two-Thirds of Next Year's Capacity Already Sold — Who's Still Shorting Memory?","type":0,"rnLink":"https://laohu8.com/RN?name=RNTheme&page=/theme/detail&rndata={\"themeId\":228f7dbe92224ab7969df668216f4d7f}&rnconfig={\"headerBarHidden\": true}","description":"Memory led Tuesday's rally as the Nasdaq closed at a record: SanDisk +6.82% to $1,887.04, Micron +5.00% to $1,096.16, SK Hynix +3.45% to $195.37. Rosenblatt started SanDisk at Buy, target $2,400, and about two-thirds of its next-year capacity is already contracted; Bernstein puts Samsung's Q3 HBM revenue up 72% QoQ. Bears had their own headlines: Michael Burry added to his Micron short on Acer's supply warning, and Micron's $25B Taiwan fab faces a possible strike — the stock rose anyway. Selling next year's capacity now: locking in profit, or borrowing from the upside?\n\n","image":"https://community-static.tradeup.com/news/c4d4b65daedcd952235a3f085db4d033"}],"images":[{"img":"https://community-static.tradeup.com/news/aa2a1a560d87994596523ff28d81aef8","width":"1254","height":"1254"}],"coverImages":[{"img":"https://community-static.tradeup.com/news/aa2a1a560d87994596523ff28d81aef8","width":"1254","height":"1254"}],"title":"AI’s Next Arms Race Isn’t Just in GPUs — It’s in Optical Interconnects","html":"<html><head></head><body>3.2T and 6.4T solutions are showing up at ECOC 2026. The next AI bottleneck may be shifting from raw compute to connectivity.\nAs AI clusters scale from thousands to hundreds of thousands of GPUs, one problem is becoming increasingly important:\nHow do you move massive amounts of data between those GPUs fast enough — and without burning too much power?\nThat is exactly why optical interconnects are becoming a bigger part of the AI infrastructure story.\nAt ECOC 2026, Coherent showcased a 3.2T OSFP optical module as well as a 6.4T NPO optical engine designed for next-generation AI scale-up and scale-out networks.\n<img data-align=\"center\" src=\"https://community-static.tradeup.com/news/aa2a1a560d87994596523ff28d81aef8\" tg-width=\"1254\" tg-height=\"1254\">\nThe direction is becoming clear:\n800G → 1.6T → 3.2T\nAnd the upgrade is not just about higher headline speeds.\nThe bigger shift is that optics are moving closer to the chip.\nAs GPU density rises, traditional electrical interconnects face growing limits in bandwidth, distance and power efficiency. That is pushing the industry toward architectures such as NPO and CPO, where optical components sit much closer to switches and compute chips.\nMarvell is also showcasing 400G/lane PAM4 technology built on 2nm, targeting future 3.2T interconnects, while 1.6T solutions are moving closer to broader deployment.\nThis matters because AI infrastructure spending is expanding beyond GPUs.\nThe first bottleneck was compute.\nThen came HBM.\nThen power.\nNow, networking is increasingly part of the constraint.\nA powerful GPU is only as useful as the system around it. If data cannot move efficiently between accelerators, expensive compute ends up waiting.\nThat means AI capex is gradually spreading into:\noptical modules, DSPs, lasers, silicon photonics, switches, CPO/NPO and data-center interconnects.\n\n<h3 style=\"text-align: left;\">Tiger View</h3>\nThe key takeaway from ECOC 2026 is not that 6.4T is suddenly ready for mass deployment.\nMany of these technologies are still in demonstration or validation stages.\nWhat matters more is the pace of innovation.\nOptical networking is evolving quickly from 800G to 1.6T and toward 3.2T, while the architecture itself is changing from pluggable optics toward NPO and CPO.\nThe goal is simple:\nmove more data, faster, with less power.\nIf GPUs are no longer the only bottleneck in AI infrastructure, then the companies that help connect compute more efficiently could become increasingly important in the next phase of the AI buildout.\nAI’s next arms race may not only be inside the GPU.\nIt may be in the light.\n<a data-mention-id=\"COHR\" rel=\"noopener noreferrer\" class=\"teditor-mention\" data-mention-name=\"Coherent\" target=\"_blank\" href=\"https://ttm.financial/S/COHR\">$Coherent(COHR)$</a> <a data-mention-id=\"LITE\" rel=\"noopener noreferrer\" class=\"teditor-mention\" data-mention-name=\"Lumentum\" target=\"_blank\" href=\"https://ttm.financial/S/LITE\">$Lumentum(LITE)$</a> <a data-mention-id=\"MRVL\" rel=\"noopener noreferrer\" class=\"teditor-mention\" data-mention-name=\"Marvell Technology\" target=\"_blank\" href=\"https://ttm.financial/S/MRVL\">$Marvell Technology(MRVL)$</a> <a data-mention-id=\"AVGO\" rel=\"noopener noreferrer\" class=\"teditor-mention\" data-mention-name=\"Broadcom\" target=\"_blank\" href=\"https://ttm.financial/S/AVGO\">$Broadcom(AVGO)$</a> <a data-mention-id=\"NVDA\" rel=\"noopener noreferrer\" class=\"teditor-mention\" data-mention-name=\"NVIDIA\" target=\"_blank\" href=\"https://ttm.financial/S/NVDA\">$NVIDIA(NVDA)$</a>\n</body></html>","htmlText":"<html><head></head><body>3.2T and 6.4T solutions are showing up at ECOC 2026. The next AI bottleneck may be shifting from raw compute to connectivity.\nAs AI clusters scale from thousands to hundreds of thousands of GPUs, one problem is becoming increasingly important:\nHow do you move massive amounts of data between those GPUs fast enough — and without burning too much power?\nThat is exactly why optical interconnects are becoming a bigger part of the AI infrastructure story.\nAt ECOC 2026, Coherent showcased a 3.2T OSFP optical module as well as a 6.4T NPO optical engine designed for next-generation AI scale-up and scale-out networks.\n<img data-align=\"center\" src=\"https://community-static.tradeup.com/news/aa2a1a560d87994596523ff28d81aef8\" tg-width=\"1254\" tg-height=\"1254\">\nThe direction is becoming clear:\n800G → 1.6T → 3.2T\nAnd the upgrade is not just about higher headline speeds.\nThe bigger shift is that optics are moving closer to the chip.\nAs GPU density rises, traditional electrical interconnects face growing limits in bandwidth, distance and power efficiency. That is pushing the industry toward architectures such as NPO and CPO, where optical components sit much closer to switches and compute chips.\nMarvell is also showcasing 400G/lane PAM4 technology built on 2nm, targeting future 3.2T interconnects, while 1.6T solutions are moving closer to broader deployment.\nThis matters because AI infrastructure spending is expanding beyond GPUs.\nThe first bottleneck was compute.\nThen came HBM.\nThen power.\nNow, networking is increasingly part of the constraint.\nA powerful GPU is only as useful as the system around it. If data cannot move efficiently between accelerators, expensive compute ends up waiting.\nThat means AI capex is gradually spreading into:\noptical modules, DSPs, lasers, silicon photonics, switches, CPO/NPO and data-center interconnects.\n\n<h3 style=\"text-align: left;\">Tiger View</h3>\nThe key takeaway from ECOC 2026 is not that 6.4T is suddenly ready for mass deployment.\nMany of these technologies are still in demonstration or validation stages.\nWhat matters more is the pace of innovation.\nOptical networking is evolving quickly from 800G to 1.6T and toward 3.2T, while the architecture itself is changing from pluggable optics toward NPO and CPO.\nThe goal is simple:\nmove more data, faster, with less power.\nIf GPUs are no longer the only bottleneck in AI infrastructure, then the companies that help connect compute more efficiently could become increasingly important in the next phase of the AI buildout.\nAI’s next arms race may not only be inside the GPU.\nIt may be in the light.\n<a data-mention-id=\"COHR\" rel=\"noopener noreferrer\" class=\"teditor-mention\" data-mention-name=\"Coherent\" target=\"_blank\" href=\"https://ttm.financial/S/COHR\">$Coherent(COHR)$</a> <a data-mention-id=\"LITE\" rel=\"noopener noreferrer\" class=\"teditor-mention\" data-mention-name=\"Lumentum\" target=\"_blank\" href=\"https://ttm.financial/S/LITE\">$Lumentum(LITE)$</a> <a data-mention-id=\"MRVL\" rel=\"noopener noreferrer\" class=\"teditor-mention\" data-mention-name=\"Marvell Technology\" target=\"_blank\" href=\"https://ttm.financial/S/MRVL\">$Marvell Technology(MRVL)$</a> <a data-mention-id=\"AVGO\" rel=\"noopener noreferrer\" class=\"teditor-mention\" data-mention-name=\"Broadcom\" target=\"_blank\" href=\"https://ttm.financial/S/AVGO\">$Broadcom(AVGO)$</a> <a data-mention-id=\"NVDA\" rel=\"noopener noreferrer\" class=\"teditor-mention\" data-mention-name=\"NVIDIA\" target=\"_blank\" href=\"https://ttm.financial/S/NVDA\">$NVIDIA(NVDA)$</a>\n</body></html>","text":"3.2T and 6.4T solutions are showing up at ECOC 2026. The next AI bottleneck may be shifting from raw compute to connectivity. As AI clusters scale from thousands to hundreds of thousands of GPUs, one problem is becoming increasingly important: How do you move massive amounts of data between those GPUs fast enough — and without burning too much power? That is exactly why optical interconnects are becoming a bigger part of the AI infrastructure story. At ECOC 2026, Coherent showcased a 3.2T OSFP optical module as well as a 6.4T NPO optical engine designed for next-generation AI scale-up and scale-out networks. The direction is becoming clear: 800G → 1.6T → 3.2T And the upgrade is not just about higher headline speeds. The bigger shift is that optics are moving closer to the chip. As GPU density rises, traditional electrical interconnects face growing limits in bandwidth, distance and power efficiency. That is pushing the industry toward architectures such as NPO and CPO, where optical components sit much closer to switches and compute chips. Marvell is also showcasing 400G/lane PAM4 technology built on 2nm, targeting future 3.2T interconnects, while 1.6T solutions are moving closer to broader deployment. This matters because AI infrastructure spending is expanding beyond GPUs. The first bottleneck was compute. Then came HBM. Then power. Now, networking is increasingly part of the constraint. A powerful GPU is only as useful as the system around it. If data cannot move efficiently between accelerators, expensive compute ends up waiting. That means AI capex is gradually spreading into: optical modules, DSPs, lasers, silicon photonics, switches, CPO/NPO and data-center interconnects. Tiger View The key takeaway from ECOC 2026 is not that 6.4T is suddenly ready for mass deployment. Many of these technologies are still in demonstration or validation stages. What matters more is the pace of innovation. Optical networking is evolving quickly from 800G to 1.6T and toward 3.2T, while the architecture itself is changing from pluggable optics toward NPO and CPO. The goal is simple: move more data, faster, with less power. If GPUs are no longer the only bottleneck in AI infrastructure, then the companies that help connect compute more efficiently could become increasingly important in the next phase of the AI buildout. AI’s next arms race may not only be inside the GPU. It may be in the light. $Coherent(COHR)$ $Lumentum(LITE)$ $Marvell Technology(MRVL)$ $Broadcom(AVGO)$ $NVIDIA(NVDA)$","highlighted":1,"essential":2,"paper":2,"link":"https://ttm.financial/post/610131541590104","repostId":0,"isVote":1,"tweetType":1,"commentLimit":10,"symbols":["NVDA","AVGO","COHR","LITE","MRVL"],"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":2121,"optionInvolvedFlag":false,"subscribersOnly":false,"subscribersOnlyAccessible":false,"xxTargetLangEnum":"ORIG"},"isVote":1,"tweetType":1,"viewCount":247,"commentLimit":10,"likeStatus":false,"favoriteStatus":false,"reportStatus":false,"symbols":[],"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":657,"optionInvolvedFlag":false,"subscribersOnly":false,"subscribersOnlyAccessible":false,"xxTargetLangEnum":"ORIG"},"commentList":[],"isCommentEnd":true,"isTiger":false,"isWeiXinMini":false,"url":"/m/post/610164137484568"}