Secondary export-restricted market: NVIDIA H800 80GB PCIe (bare card, ASKING band)
Fields
supply/used-h800-80gb-marketNotes
ACCELERATOR-ID CORRECTION (2026-10-03, parent agent): accelerator_ids previously read nvidia-h100-pcie, which is a different part in a different market (H800 is the export-controlled China-market H100 variant with reduced NVLink and FP64). Repointed to nvidia-h800-80gb. NOTE: no nvidia-h800-80gb accelerator record exists in the repo yet -- this is a known dangling reference and should stay a warning, not be silently dropped. OBSERVATION DATE 2026-10-03. Source sup-price-picclick-ebay-asking. OBSERVED: three listings at $33,900 / $41,221.28 (twice - two different sellers running the identical Chinese-language title, so the duplicate is one price quoted twice, not two observations). Effective band $33,900-$41,221 with n_effective=2. Recorded deliberately as its own segment and NOT as an H100: the H800 is the export-control-downgraded Hopper (reduced NVLink/interconnect), it is not an H100, and folding it into an H100 band would import the interconnect restriction into the price. export_controlled is true for the underlying part under the US export-control regime; the listings themselves are third-party resale, whose legality varies by destination.
EXPORT CONTROL (2026-10-04, mechanism-level; sources spl-eccn-3a090-ccl, spl-bis-2026-01-advanced-computing-case-by-case, spl-bis-2026-05-d5-guidance): these parts classify under ECCN 3A090.a/.b, whose thresholds are 'total processing performance' (TPP) >= 4800, or TPP >= 1600 with performance density >= 5.92 (the .b band is TPP 2400-4800 with performance density 1.6-5.92, or TPP >= 1600 with performance density 3.2-5.92). The mechanism has THREE layers and all three must be cleared, not just one: (a) LICENSE FOR EXPORT FROM THE US - controlled by the January 2026 rule (91 FR 1684, effective 2026-01-15), which moved TPP < 21,000 AND total DRAM bandwidth < 6,500 GB/s parts from presumption of denial to CASE-BY-CASE REVIEW, but only for exports from the US to end users in China/Macau, and only against six exporter certifications - US supply sufficiency judged 'taking into account normal lead times', no diversion of foundry capacity away from US end users, aggregate China+Macau shipments capped at 50% of US end-use shipments, no end-user prohibition, consignee KYC procedures, and prior verification by a qualified third-party US testing lab of TPP, total DRAM bandwidth, interconnect bandwidth and copackaged DRAM capacity; (b) END-USER AND PARENT-COMPANY SCREENING under 742.6(a)(6)(iii)(A) - per BIS guidance of 2026-05-31 a licence is required where the consignee is HEADQUARTERED in Country Group D:5 or Macau OR ITS ULTIMATE PARENT is so headquartered, EVEN IF the consignee itself is outside D:5/Macau, which means the destination country alone never answers the question; (c) THE LIMITS THAT REMAIN - reexports and in-country transfers of the same chips to China/Macau remain PRESUMPTION OF DENIAL, and exports to D:5/Macau-headquartered or -parented entities remain presumption of denial wherever the buyer sits. The Jan-2025 AI Diffusion Rule was rescinded and not enforced from 2025-05-13, but that rescission did not remove (a) or (b). NOTE SPECIFIC TO H800: this part exists only as an export-control artefact - it was the interconnect-reduced H100 variant NVIDIA built to comply with the Oct-2023 thresholds, and the Oct-2023 rule closed that route. Any H800 in the market is therefore pre-rule stock or diversion, not a currently orderable SKU, which changes what its asking price means: it is a thin, non-repeatable supply, not a price you can plan around for a new deployment.
LEAD TIME: null, and deliberately so. A secondary-market LISTING has no lead time - the part either exists on the marketplace on the observation date or it does not, and the seller's dispatch time is not published and not a commitment. This is the opposite failure mode from a distributor: a backordered new part at least carries a distribution window (CDW prints '4-6+ Weeks' on some SKUs, source spl-cdw-published-lead-times-gpu-2026-10), whereas a used listing with no buyer carries nothing at all. The correct consequence for a reader is that used and auction supply cannot be scheduled around and therefore cannot carry a build plan - they are opportunistic capacity, which is the strongest argument in this slice for the renting-over-buying decision. Probe log in source spl-negative-no-oem-publishes-lead-time.