Analyst Zephyr says OpenAI's custom Jalapeño inference chip closely resembles Microsoft's Maia 200, sparking debate over TPU similarities

some info on the OpenAI inference chip from the image + reporting:

central die: ~25.78mm × ~32.23mm ≈ 831 mm²

- top-left/top-right are probably spacers/dummy silicon - top-middle might be networking / scale-up IO

Overall it looks very similar to Maia 200 / TPU8t, which means: - near-reticle inference ASIC - 6× HBM3E 12-high = 216GB - ~7.1–7.4TB/s bandwidth - ~10PFLOPS FP4