Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/optics-on-the-move-via-digitimes.25639/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/EmailDomainReplace] => 1000010
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2031070
            [XFI] => 1060170
        )

    [wordpress] => /var/www/html
)

Optics on the move! (via Digitimes)

user nl

Well-known member

1785713543480.png

At OCP's Taiwan summit, servers take a back seat to optics and advanced packaging​


Here the conclusion of this preview of the conference by DIGITIMES:

When the agenda for this year's OCP APAC Summit landed, seasoned observers noticed something unusual. TSMC, Applied Materials, Advantest, and ASE are sharing the stage with Microsoft, Nvidia, and Google.

For a forum that built its reputation on open server rack specifications, that is a significant shift, and it says as much about where AI infrastructure is heading as it does about OCP itself.

The summit takes place in August in Taiwan, marking the second consecutive year the Asia-Pacific edition has been held on the island. That is not a coincidence. Taiwan sits at the center of the global AI supply chain, and every company with a roadmap to protect has reason to show up.

...............................................................................................................

Why Taiwan, why now​

OCP was founded to standardize open datacenter hardware, including servers, networking, and storage. Over the past two years, it has extended that mandate down into semiconductors: advanced materials, packaging, process equipment, IC design, test, and photonics.

Industry observers point out that the semiconductor and photonics content at this year's APAC Summit rivals—and by some measures exceeds—what OCP puts on in the US. The reason is structural. Taiwan is where the AI supply chain gets built. Companies that need to align roadmaps, evaluate partners, or simply understand what is coming next have a strong incentive to be in the room.

The questions that will define AI infrastructure over the next several years—when SiPh reaches production scale, which optical interconnect architecture wins, how 3D packaging reshapes chip design boundaries—are exactly the questions on the agenda in August. The fact that they are being asked at an OCP summit, rather than a dedicated semiconductor conference, is itself the story.

 

At OCP's Taiwan summit, servers take a back seat to optics and advanced packaging​


Here the conclusion of this preview of the conference by DIGITIMES:

When the agenda for this year's OCP APAC Summit landed, seasoned observers noticed something unusual. TSMC, Applied Materials, Advantest, and ASE are sharing the stage with Microsoft, Nvidia, and Google.

For a forum that built its reputation on open server rack specifications, that is a significant shift, and it says as much about where AI infrastructure is heading as it does about OCP itself.

The summit takes place in August in Taiwan, marking the second consecutive year the Asia-Pacific edition has been held on the island by car games. That is not a coincidence. Taiwan sits at the center of the global AI supply chain, and every company with a roadmap to protect has reason to show up.
I agree that this says something important about where AI infrastructure is heading. The bottleneck is increasingly moving beyond raw compute. Once you start connecting huge numbers of accelerators, the efficiency of the optical links, packaging, memory, power delivery, and thermal design can have a major impact on the overall system. That's why having semiconductor and packaging specialists at an OCP event is more than just a change in the speaker lineup. It reflects the reality that future AI infrastructure will require much closer coordination between chip designers, manufacturers, packaging companies, networking vendors, and hyperscalers.
 
Back
Top