Micron has ceased production of its 2 GB GDDR7 memory chips, signalling a shift towards 4 GB and 6 GB options that could revolutionise VRAM capacity in future graphics cards, potentially simplifying design and reducing costs amid industry anticipation for the new high-density memory solutions.
Micron has ended production of its 2 GB GDDR7 memory chips, while the industry is already looking towards denser 4 GB and 6 GB parts that could arrive in 2027 or 2028, according to reports based on information circulating among memory suppliers and hardware leakers. If that timetable holds, future graphics cards could gain more video memory without requiring wider buses or major changes to board layouts.
At present, the market remains constrained by limited availability of high-density GDDR7. NVIDIA’s GeForce RTX 5090, for example, uses 2 GB chips, while 3 GB modules have seen narrower adoption because they are expensive and do not always fit neatly into existing GPU memory configurations. The result is that card makers often have to balance capacity, cost and bus width rather than simply increasing VRAM.
The reported move towards 4 GB and 6 GB chips would change that calculation. A graphics card with a 128-bit memory interface could, in theory, move from 8 GB of VRAM with four 2 GB chips to 16 GB with four 4 GB chips, or 24 GB with four 6 GB chips, without altering the number of packages or the bus width. The same logic applies to 256-bit designs, where eight 2 GB chips provide 16 GB, while eight 4 GB chips would yield 32 GB and eight 6 GB chips would take that to 48 GB.
That matters most for entry-level and mid-range cards, where 128-bit buses are common and memory capacity is often a limiting factor. Higher-density chips could let manufacturers offer more VRAM while keeping the overall memory subsystem simpler, and in some cases potentially cheaper, than redesigning the board for a broader interface. The approach would also be useful for professional accelerators and workstations, where large memory pools are increasingly important for artificial intelligence and other data-heavy workloads.
Micron’s withdrawal from the 2 GB segment does not appear to create an immediate supply crisis for current GeForce RTX 50 boards. The company only began shipping GDDR7 to NVIDIA in meaningful volumes this year and is not the dominant supplier for the range, with Samsung and SK Hynix still covering much of demand. Micron is also still understood to be producing 3 GB GDDR7 parts, although those higher-capacity chips cannot simply be dropped into existing designs without a board redesign.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
