SanDisk2026-08-14 13:21:06Citrini analyst questions SanDisk’s HBF demo, says HBM performance may have been understatedCitrini analyst Zephyr challenged SanDisk’s HBF-versus-HBM comparison shown at its investor day, arguing that the setup may have understated HBM’s real capabilities. In his post, he said SanDisk assigned the same total bandwidth of 12.8 TB/s to both HBF and HBM, which works out to 1.6 TB/s per stack, and used the Qwen3-480B-A35B model running in bfloat16 for the demonstration. Zephyr said current inference workloads more commonly use FP4 or FP8, putting the model’s memory requirement at roughly 240 GB to 480 GB depending on quantization. He also said SanDisk fixed HBM capacity at 192 GB per GPU, while a 16-layer HBM4E setup with eight stacks can reach 512 GB and deliver about 32 TB/s of bandwidth, around three times the level used in the company’s demo. Based on that, Zephyr argued that SanDisk’s result showing HBF needing only one to four GPUs while HBM needed eight does not adequately reflect the performance of higher-spec HBM configurations.1470
Kimi K32026-07-17 11:11:01Citrini analyst says Kimi K3 could squeeze top AI model margins and benefit AI infrastructure playsCitrini analyst Zephyr commented on the newly released Kimi K3 model on July 17, saying it could compress profit margins for leading AI model companies. In Zephyr’s view, OpenAI’s Sol still holds a pricing edge, while Anthropic’s Opus is under heavier pressure. The analyst also said K3 is positive for cloud service providers, Token-as-a-Service providers, and the broader AI infrastructure segment. Zephyr added that Moonshot is expected to buy large amounts of computing power as demand jumps, a development he said would directly benefit A-share companies tied to AI infrastructure. The remarks were reported by BlockBeats.1470
