A display and a memory spec, both pushing at ceilings. ASUS takes third-generation QD-OLED to 500 Hz by holding resolution at QHD, and the first formal High Bandwidth Flash spec describes NAND stacks doing HBM’s job at eight times the capacity.
ASUS pushes QD-OLED to 500Hz for competitive play
ASUS listed the ROG Swift OLED PG27AQDPR on August 5: a 27-inch third-generation QD-OLED panel running QHD resolution (2560×1440) at up to 500Hz, with a claimed 0.03ms gray-to-gray response time, DisplayHDR 500 True Black certification, and 99 percent DCI-P3 coverage (HotHardware on the PG27AQDPR, VideoCardz on the listing). The monitor keeps QHD rather than chasing 4K, which is the tradeoff that makes 500Hz achievable at all right now: fewer pixels to refresh means more headroom for panel and drive electronics to hit an aggressive rate, the same reason the highest refresh OLED panels on the market keep landing at 1440p rather than 4K. It is built for competitive shooters and esports players who read motion clarity and input latency ahead of resolution, not a general-purpose upgrade.
ASUS has not announced pricing or a release date. The closest existing sibling, the ROG Strix XG27AQDPG, carries an $849 MSRP and has been seen discounted to $699, which gives a rough anchor but nothing more; treat any specific number for the PG27AQDPR as a guess until ASUS confirms one.
A NAND-based alternative to HBM gets its first real spec
Sandisk and SK hynix published the first formal High Bandwidth Flash specification through the Open Compute Project, with Google DeepMind and Tenstorrent named as ecosystem partners. The spec defines three bandwidth tiers running from roughly 400 GB/s up to 3 TB/s, with individual 8-high or 16-high 3D NAND stacks reaching up to 512GB, against the roughly 64GB ceiling on a single HBM4 stack. That capacity gap is the entire pitch: HBF targets AI inference workloads, particularly mixture-of-experts models, that need a large pool of memory close to the compute die more than they need HBM-class latency on every access.
None of this reaches a graphics card soon, or possibly ever in a consumer form. The spec targets data-center packaging, and Sandisk and SK hynix frame it as enterprise-only for now, with the UCIe chiplet integration and heterogeneous-packaging costs still without a consumer path. It's worth tracking mainly as a signal of where relief for the memory-capacity pressure behind current GDDR and HBM pricing might eventually come from, even while that relief stays architecturally distant.