SOCAMM will not be a repackaging of present {hardware} or an answer searching for an issue. Relatively, it’s addressing actual issues and is required and needed by the trade, mentioned Jim Handy, president of Goal Evaluation.
“The server processor producers and Nvidia are actually behind SOCAMM due to the truth that they [gain] a quicker interface and it will get a whole lot of reminiscence right into a small space with just a little bit decrease energy consumption,” he mentioned.
As beforehand famous, SOCAMM2 makes use of stacked reminiscence, which may be costlier to fabricate, however Useful says that’s not the case. “[Memory vendors] promote totally different stack configurations for a similar value as these DRAM guys, and they’re utilizing the identical packaging expertise that the NAND flash guys do, so I wouldn’t count on there to be any noticeable value distinction,” he mentioned.
SK Hynix, one other main reminiscence producer available on the market, has additionally introduced plans to help SOCAMM2 reminiscence however has not given any launch particulars and is believed to be behind Micron and Samsung. SOCAMM2 is anticipated to launch across the second quarter of 2026 when Nvidia launches its Vera Rubin platform.
