HBM4 Archives | BonTech Labs

Two SK Hynix HBM4 memory chips are shown; one chip displays the label “HBM4” while the other is flipped to reveal its gold connector side, highlighting advanced packaging. Both are set on a plain white background.

HBM4 and Packaging in 2026: The Bottleneck No One Wants To Own

November 14, 2025 Gavin Bonshor 0

2026 will not be won by a shinier compute tile, but it’ll be won by the teams that can bolt enough healthy HBM onto a […]

A close-up and internal view of a high-performance computer server, showcasing stacked components and detailed circuitry—featuring the NVIDIA Rubin CPX for disaggregated inference, with a highlighted section zooming in on chips and hardware inside the server.

NVIDIA Rubin CPX Explained: Disaggregated Inference And The Cost Of Million-Token Context

November 13, 2025 Gavin Bonshor 0

NVIDIA’s Rubin platform splits long-context prefill from token decode. Rubin CPX handles the compute-heavy front half, standard Rubin handles bandwidth-heavy generation. The NVL144 CPX rack […]

A close-up of a Micron HBM4 memory chip, labeled “micron HBM4,” sits beside an exposed view showing intricate circuitry on a gold-colored substrate—essential technology for AI accelerators—set against a black background.

Micron’s HBM4 clears 2.8 TB/s with samples shipping — HBM4E custom base die points to bespoke AI accelerators

October 3, 2025 Gavin Bonshor 0

Micron says its next-gen HBM4 already exceeds 2.8 TB/s per stack with >11 Gb/s pin speeds, and that it’s begun shipping customer samples. The company […]