Nvidia is reportedly preparing a lower-cost AI chip for China, with a June launch planned, according to Reuters. The chip is based on the Blackwell architecture and the RTX Pro 6000D design, but it appears to be built around a narrower goal: staying useful for the Chinese market while fitting within U.S. export rules that restrict sales of high-end chips to China.
A lower-cost Blackwell option for China
The reported product would give Nvidia a cheaper GPU for one of the most important AI hardware markets. Reuters says the new chip is expected to cost $6,500 to $8,000. That places it below the H20 model, which is listed in the source at $10,000 to $12,000.
The pricing matters because it signals that the product is not simply another high-end Blackwell chip routed into China. It is described as a lower-cost GPU based on Blackwell architecture and the RTX Pro 6000D design. In plain terms, Nvidia appears to be shaping a product that can still carry the Blackwell name while using a configuration designed for different constraints.
The source does not describe performance targets, customer names, or exact technical specifications beyond the architecture, design basis, packaging, memory, price range, and launch timing. That means the clearest takeaway is not a benchmark claim. It is that Nvidia is reportedly adjusting its AI chip lineup for China around cost, production practicality, and export-rule limits.
Why the design choices matter
Three people familiar with the plan told Reuters that the chip avoids advanced packaging from TSMC and uses standard GDDR7 memory. Those details point to a product designed to be easier to produce than a more complex high-end accelerator.
Advanced packaging is mentioned in the source as something this GPU avoids. The reported use of standard GDDR7 memory also suggests a design that relies on more conventional components. The article does not say that these choices alone define the chip, but it does say they make it easier to produce.
That production point is central. A chip aimed at China under U.S. export rules has to solve more than one problem. It must fit a regulatory environment that limits high-end chip sales, while also being manufacturable at a price point below the H20 model. Based on the source, Nvidia’s reported answer is a Blackwell-based GPU with simpler production requirements and a lower expected cost.
The export-rule context
Reuters frames the move as a response to U.S. export rules that limit the sale of high-end chips to China. That context explains why the reported chip is not described only as a cheaper product. It is also a market-specific product shaped by restrictions on what Nvidia can sell.
For Nvidia, China remains important enough to justify a tailored AI chip strategy, at least according to the reported plan. But the company cannot simply treat the market as identical to unrestricted markets. The source makes clear that the new GPU is connected to export limits, which means the product’s design and pricing have to be understood through that lens.
The chip’s expected price range also fits the broader picture. A product priced below the H20 model could appeal to buyers looking for Nvidia AI hardware under a different cost structure. At the same time, the lower price and production-oriented design choices may reflect the realities of building a compliant product for a constrained market.
What is known, and what is not
The confirmed details in the source are limited, but specific. The reported chip is:
- Planned for launch in June, according to Reuters.
- Based on Blackwell architecture and the RTX Pro 6000D design.
- Expected to cost $6,500 to $8,000.
- Cheaper than the H20 model, listed at $10,000 to $12,000.
- Designed without advanced packaging from TSMC.
- Built with standard GDDR7 memory.
- Described as easier to produce.
- Positioned as a response to U.S. export rules limiting sales of high-end chips to China.
Just as important are the details not included in the source. There are no named customers, no official Nvidia statement, no performance comparison, and no full specification sheet in the provided article. There is also no claim that the chip replaces the H20 model, only that it is expected to cost less.
That distinction matters for understanding the report responsibly. The new GPU, as described, is a lower-cost Blackwell-based AI chip for China. It is not presented as Nvidia’s most powerful chip, and the source does not support claims about how it will perform against other AI accelerators.
The bigger signal from Nvidia
The reported launch shows how AI hardware strategy can be shaped by regulation as much as by raw performance. Nvidia is said to be preparing a chip that keeps a connection to the Blackwell architecture while making different choices around memory, packaging, production, and price.
For readers watching the AI chip market, the important point is the pattern. When export rules limit the sale of high-end chips to China, companies may respond by designing products specifically for that environment. Reuters reports that Nvidia’s lower-cost Blackwell AI chip is one such response.
If the June launch proceeds as reported, Nvidia would add another China-focused option to its AI GPU lineup. Based on the source, the product’s defining features are not dramatic public claims, but practical constraints: a lower price than H20, standard GDDR7 memory, no advanced packaging from TSMC, and a design intended to be easier to produce.