Short answer: Reuters reported on August 19, 2025, that Nvidia was developing a Blackwell-based accelerator tentatively called B30A for China. The reported chip was intended to outperform Nvidia’s H20 while staying within U.S. export-control limits. Nvidia has not publicly confirmed a B30A launch, final specifications, volume production, customer shipments, or U.S. export approval. A later U.S. policy change created a separate, conditional licensing route for Nvidia’s existing H200, so H200 approvals should not be treated as proof that B30A reached the market.
What Nvidia was reported to be developing
Reuters said people briefed on Nvidia’s plans described a new China-focused accelerator, tentatively named B30A, using the Blackwell architecture. The report said the design was intended for Chinese cloud and technology companies and would be more capable than the H20, Nvidia’s earlier China-compliant accelerator. Specifications were still subject to change when the report appeared. Reuters’ August 19, 2025 report was based on anonymous sources, not a public Nvidia product announcement.
Later technology coverage used other labels, including B20A for a China-oriented Blackwell training product and RTX 6000D for an inference-focused product. Those names may describe separate products, evolving internal labels, or projects at different stages; the available reporting does not establish that they are all the same chip. Tom’s Hardware summarized those reported names and roles.
“Release” can mean several different things
A reported design plan is not the same as a commercial launch. It could refer to completing a design, sampling it with customers, preparing manufacturing, or shipping finished systems. No reviewed public Nvidia material confirms which, if any, of those milestones B30A reached.
#1 Best Overall
- AI Performance: 767 AI TOPS
- OC mode: 2632 MHz (OC mode)/ 2602 MHz (Default mode)
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Axial-tech fan design features a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- A 2.5-slot design maximizes compatibility and cooling efficiency for superior performance in small chassis
Why Nvidia needed a China-specific accelerator
U.S. export controls limit the sale of advanced computing products to China using technical thresholds and end-user rules. Nvidia’s earlier H20 was designed for the Chinese market after restrictions made higher-end products difficult or impossible to export.
Nvidia disclosed that the U.S. government told it in April 2025 that a license was required to export H20 integrated circuits, and other circuits meeting specified H20-related thresholds, to China, Hong Kong, Macau and certain other destinations. The company’s filing is available in its fiscal 2026 third-quarter filing.
The commercial challenge is a moving target: Nvidia must offer enough performance to attract buyers while satisfying U.S. rules, obtaining any required licenses, securing manufacturing capacity and meeting Chinese procurement requirements. A chip that complies when designed can become restricted after a rule change, and technical compliance does not guarantee an export license or a Chinese purchase order.
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5070 Ti
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
How B30A, H20 and H200 differ
| Product | Role in the story | Export status established by available sources |
|---|---|---|
| H20 | Earlier Nvidia accelerator modified for China | Subject to new U.S. licensing requirements in 2025. Nvidia reported no sales to China-based customers in the quarter ended July 27, 2025. |
| B30A | Reported Blackwell-based China-specific project | No public Nvidia confirmation of a launch, final specifications, volume production, customer shipments or U.S. approval. |
| H200 | Existing, more powerful Nvidia data-center accelerator | The Bureau of Industry and Security moved applications to case-by-case review in January 2026; later approvals and China-related production were reported. |
Nvidia’s fiscal second-quarter 2026 results said it had no H20 sales to China-based customers for the quarter ended July 27, 2025, while also describing significant financial effects from the licensing requirement. Nvidia’s results release provides the company’s account.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Performance claims need a metric
The defensible comparison is the one Reuters attributed to its sources: B30A was expected to be more powerful than H20. Some secondary accounts suggested a fraction of the compute capability of a dual-die flagship Blackwell configuration, and some repeated claims that it could exceed H100 performance. Those are reported estimates, not Nvidia specifications or public benchmarks.
AI-accelerator performance depends on precision, memory capacity and bandwidth, interconnect, software, model architecture, cluster size, power and cooling. “Faster” or “more powerful” is therefore incomplete without identifying the workload and measurement.
Rank #3
- Powered by the NVIDIA Blackwell architecture and DLSS 4. System Requirements: Minimum 850W PSU with 16-pin 12V-2x6 (12VHPWR) connector required. Verify before purchasing.
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability. Compatibility: 348mm (13.7") length, 3.6 slots, 4.3 lbs. Confirm case clearance and slot spacing. GPU bracket included.
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.6-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
What changed in 2026: the H200 route
On January 13, 2026, BIS announced that license applications for Nvidia H200, AMD MI325X and similar products would receive case-by-case review. The policy described conditions including preserving semiconductor production capacity for U.S. customers, purchaser compliance and screening procedures, and independent testing in the United States for performance and security. BIS published the policy announcement.
In March 2026, Jensen Huang said Nvidia was restarting H200 production for approved China-related customers, according to Axios. That development gives Nvidia a documented route to some Chinese demand, but it is not evidence that B30A was approved, launched or abandoned.
Recommended Free Tools
Nvidia’s fiscal 2026 fourth-quarter release also excluded China data-center compute revenue from its fiscal 2027 first-quarter outlook, underscoring how uncertain the market remained in the company’s official guidance. Nvidia’s release contains that outlook.
Rank #4
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5060
- Integrated with 8GB GDDR7 128bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
Why approval in Washington may not be enough
Chinese authorities and buyers can impose constraints of their own. Reuters reported that China urged some firms to avoid H20 chips, particularly for government or national-security work, while domestic alternatives such as Huawei’s Ascend products were promoted. Reuters’ report illustrates why a U.S.-approved Nvidia product may still face limited Chinese demand.
The policy conflict has three objectives:
- Washington: restrict advanced computing that it believes could support military modernization, surveillance, cyber operations or strategic AI development. BIS rules covering the Commerce Control List and Chinese military end-use or end-user restrictions are set out in Part 738 and Part 744.
- Nvidia: preserve access to a major AI-computing market and prevent customers from permanently shifting to domestic alternatives.
- China: obtain useful computing capacity while reducing dependence on U.S. suppliers and accelerating domestic chip development.
What remains unknown about B30A
- Whether B30A was the final product name.
- Final compute, memory, interconnect, power and software specifications.
- Price, manufacturing status and production volume.
- Whether any customer received samples or shipments.
- Whether a U.S. license was requested or granted.
- Whether Chinese regulators or state-linked buyers approved its use.
- Whether the project was canceled, redesigned, merged with another Blackwell product or superseded by H200 availability.
What this means for buyers and investors
H200 and Blackwell systems are enterprise products normally obtained through Nvidia, authorized server manufacturers, systems integrators or cloud providers—not ordinary retail purchases. Nvidia’s H200 product page and Blackwell platform page describe the families, but a general product page does not establish China export eligibility.
Organizations that need occasional access may rent GPUs from providers such as AWS, Microsoft Azure, Google Cloud, Oracle Cloud or CoreWeave. Cloud availability does not bypass export controls, customer screening, sanctions, data-residency rules or Chinese procurement restrictions.
Best Value
- Powered by the NVIDIA Blackwell architecture and DLSS 4 OC mode: 2640MHz/Default mode: 2610MHz (Boost Clock)
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.125-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
AMD’s Instinct MI325X is a competing accelerator specifically named in BIS’s January 2026 policy. Compatibility, software support, migration cost and availability must be assessed for the buyer’s models; no equivalent performance should be assumed without workload-specific benchmarks.
Bottom line: a credible report, not a confirmed product
The strongest current conclusion is narrow: Nvidia was credibly reported in August 2025 to be developing a more capable, export-conscious Blackwell accelerator for China, tentatively called B30A. There is no public evidence in the reviewed sources that Nvidia formally launched it or that it entered approved, volume commercial shipment. The later H200 licensing policy and reported production restart are separate developments. Nvidia’s China strategy now depends on changing U.S. rules, Chinese procurement preferences, supply-chain capacity and whether customers will accept an export-compliant product at all.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
