Leading memory manufacturers are currently working on the next generation of memory standard – DDR6 – which is currently on track for a commercial release in 2027. Taiwan-based outlet Commercial Times reported that companies like Samsung, Micron and SK Hynix are collaborating with Intel and AMD on interface testing, with platform validation expected to begin by next year.
DDR6 On The Way
DDR6 is expected to bring significant performance improvements over DDR5, with standard speeds starting at 8,800MT/s and scaling up to 17,600MT/s; some overclocked modules are anticipated to push beyond that, potentially reaching speeds of up to 21,000MT/s. Of course, both Intel and AMD will eventually support DDR6 in their future CPU platforms, and it is reported the process begins as soon as 2026, starting from AI and HPC segments.
One of the key architectural changes in DDR6 is a shift to a four-channel design using 24-bit sub-channels, replacing the dual 32-bit layout of DDR5. This change is intended to enhance data throughput and bandwidth efficiency, but it also increases the complexity of I/O design and demands better signal integrity. It’s also likely that the base voltage will be lowered to just 1.0V or even lower, which improves power efficiency as well.
DDR6 will likely move away from UDIMM, however – the current form factor found in PCs today is starting to hit its practical limits (even more so in SO-DIMM formats), so memory makers are looking to make the switch to the new CAMM2 standard to get around this. This new form factor allows for higher capacity per module and improves on thermal dissipation capabilities, which G.Skill showcased in Computex 2025 earlier this year with its 64GB DDR5-10000 CAMM2 module.
Pokdepinion: I imagine this means increased build cost, which means pricier modules – similar to PCIe 5.0 motherboards.
