Skip to main content

NVIDIA H20 (2024)

Product Overview​

NVIDIA H20 is the export-compliant AI accelerator NVIDIA launched for the China market, based on the Hopper architecture (same architecture as the H100/H200), officially released in early 2024.

Restricted by US export controls (the October 2023 rules), H20's compute was deliberately cut to comply:

  • FP16: 148 TFLOPS (~15% of the H100)
  • FP8: 296 TFLOPS (~15% of the H100)
  • Memory: 96GB HBM3 (4.0 TB/s bandwidth)

H20's positioning: keep NVIDIA's ecosystem advantages as much as possible while staying compliant, competing with domestic AI chips (Hygon DCU, Kunlunxin, MetaX, etc.).

Core Specifications​

ParameterValue
ArchitectureHopper (export-compliance cut-down)
ProcessTSMC 4N (4nm)
FP8296 TFLOPS
FP16148 TFLOPS
TF3274 TFLOPS
FP3224 TFLOPS
INT8296 TOPS
Memory Capacity96GB HBM3
Memory Bandwidth4.0 TB/s
InterconnectNVLink 900 GB/s (downgraded)
TDP400W
ReleaseEarly 2024
DiscontinuedJuly 2025 (NVIDIA notified channel partners)

Export Compliance Background​

US Export Control Timeline​

DateEvent
2022-10US BIS issues export controls restricting A100/H100 exports to China
2023-10New rules tighten further, adding a "performance density" threshold
2023-11NVIDIA launches H20 / L20 / L2 (compliant versions)
2024-02H20 begins shipping to Chinese customers
2025-07NVIDIA notifies channel partners that H20 will be discontinued (export controls tighten again)

Compliance-Limited Parameters​

ParameterCompliance ThresholdH20 Actual
FP16 compute< 300 TFLOPS148 TFLOPS
Performance density< a certain thresholdCompliant
Memory bandwidthNo explicit limit4.0 TB/s (uncut)

Strategy: NVIDIA chose to keep memory bandwidth (important for LLM inference) while cutting compute sharply, staying competitive within compliance.

Comparison with Domestic Chips​

H20 vs Hygon DCU K100​

MetricNVIDIA H20Hygon DCU K100Difference
FP16148 TFLOPS192 TFLOPSK100 +30%
Memory96GB HBM364GB HBM3H20 +50%
Bandwidth4.0 TB/s3.2 TB/sH20 +25%
EcosystemCUDA (complete)ROCm (compatible)H20 advantage
SupplyDiscontinued in 2025Stable supplyK100 advantage
Price~$25k (est.)~$15k (est.)K100 cheaper

H20 vs Kunlunxin P800​

MetricNVIDIA H20Kunlunxin P800Difference
FP16148 TFLOPS345 TFLOPSP800 2.3x
Memory96GB HBM332GB HBM3H20 +200%
Bandwidth4.0 TB/sNot disclosedH20 advantage
EcosystemCUDABaidu PaddleEach has merits

Conclusion: H20 still leads in memory capacity and bandwidth, but its compute has been overtaken by domestic chips. With H20 discontinued in 2025, domestic substitution is accelerating.

Use Cases​

  • ✅ Transition period for NVIDIA ecosystem migration (CUDA code needs no changes)
  • ✅ LLM inference (96GB memory + 4.0 TB/s bandwidth advantage)
  • ✅ Short-term projects (still procurable before July 2025)
  • ❌ Long-term training clusters (discontinued in 2025, supply chain risk)
  • ❌ Cost-effectiveness first (domestic chips offer higher compute at lower prices)
  • ❌ Self-reliance requirements (the US government can tighten export controls at any time)

Discontinuation Impact (July 2025)​

In July 2025, NVIDIA notified channel partners that H20 would be discontinued, because:

  1. US export controls could tighten further
  2. H20 margins were squeezed (deliberately cut compute)
  3. NVIDIA prioritized capacity for H200/B200

Impact on the China market:

  • Customers who bought H20: no short-term impact; repairs/expansion difficult long-term
  • Customers who did not: accelerated shift to domestic chips (Hygon, Kunlunxin, MetaX, etc.)
  • Domestic substitution timeline: brought forward 1-2 years

Vendor Information​

ParameterValue
CompanyNVIDIA Corporation
ArchitectureHopper (export-compliance cut-down)
ReleaseEarly 2024
DiscontinuedJuly 2025 (channel partners notified)
China market positioningExport-compliant version replacing the H100/A100
CompetitorsHygon DCU, Kunlunxin, MetaX, Biren, etc.