NVIDIA chief government Jensen Huang is fretting. That is evident within the lack of alternative to do enterprise in China, one of many world’s main AI markets with a $50 billion alternative. In current months, Huawei Applied sciences, NVIDIA’s Chinese language counterpart, has gained inNVIDIA’s absence and isbuilding a sophisticated chip manufacturing facility to ship a number of components for the AI provide chain.
Even because the US chip maker maintains a 90% share within the world GPU market, the potential lack of alternative, roughly price $8 billion USD for the upcoming quarter, a sixth of its Q1 income, has riled them up.
US President Donald Trump maintained the export restrictions on Chips, amid repeated warnings from business leaders that it may threaten the nation’s AI dominance. Huang referred to as it a significant strategic blow to the US and mentioned the restrictions have spurred China’s innovation and scale.
“Shielding Chinese language chipmakers from U.S. competitors solely strengthens them overseas and weakens America’s place,” Huang mentioned in the course of the firm’s earnings name on Wednesday, calling the East Asian nation a springboard to world success.
“With half of the world’s AI researchers based mostly there, the platform that wins China is positioned to steer globally,” he mentioned. “China’s AI strikes on with or with out U.S. chips. It has to compute to coach and deploy superior fashions.”
Huang mentioned the nation’s assumption that China can’t make AI chips was flawed, and contemplating China’s monumental manufacturing capabilities, it has proved them improper now.
On April 9, 2025, the US authorities knowledgeable NVIDIA {that a} license is required to export its H20 merchandise to the Chinese language market. On account of these new necessities, NVIDIA incurred a $4.5 billion cost within the first quarter of fiscal 2026.
The corporate reported $4.6 billion in H20 product gross sales within the quarter previous to the licensing requirement.
NVIDIA CFO Colette Kress mentioned in the course of the earnings name that the corporate continues to be evaluating its restricted choices to produce knowledge centre compute merchandise to China, compliant with the US authorities’s revised export management guidelines.
She added that within the knowledge centre, the corporate anticipates the continued ramp of Blackwell to be partially offset by a decline in China income. “Our outlook displays a loss in H20 income of roughly $8 billion for the second quarter,” Kress mentioned.
In keeping with Taiwanese outlet Digitimes, which cited sources inside the provide chain, NVIDIA and AMD are making ready to launch new AI GPUs for the Chinese language market to adjust to US export restrictions on superior semiconductor know-how.
NVIDIA is reportedly set to introduce a scaled-down AI GPU identified internally because the “B20”, whereas AMD plans to supply its new Radeon AI PRO R9700 workstation GPU to satisfy native AI computing wants. Each corporations are anticipated to start shipments in China by July.
Individually, Reuters lately reported that NVIDIA is creating a lower-cost AI chip based mostly on its Blackwell structure for the Chinese language market. The chip’s projected value vary is $6,500 to $8,000, which might place it beneath the value of its H20 GPU, which at the moment sells for between $10,000 and $12,000.
The Rise of Huawei
Notably, in line with a Reuters report, Huawei is ready to start mass shipments of its Ascend 910C AI chip to Chinese language prospects in Could. The Ascend 910C is an upgraded GPU that mixes two 910B processors right into a single bundle, successfully doubling computing energy and reminiscence capability.
In the meantime, Huawei is making ready to check its new AI chip, the Ascend 910D, in China, whereas it has additionally launched the Ascend 920 AI chip.
In keeping with DigiTimes Asia, the chip is slated to enter mass manufacturing within the second half of 2025. Trade consultants imagine the Ascend 920 may very well be a viable various to NVIDIA’s H20 GPUs.
In the meantime, Cambricon Applied sciences can also be gaining prominence in China. In a current report, Cambricon mentioned it had efficiently optimised Qwen3 to run effectively on its GPUs. This optimisation was pushed by demand from AI builders within the Philippines searching for China-made chips.
In the meantime, Trump additionally warned in opposition to utilizing Chinese language AI chips, resulting in felony penalties.
$44.1 Billion in Income Regardless of Hit
Regardless of the setback in China, NVIDIA reported $44.1 billion in income in Q1FY26, up 12% from the earlier quarter and 69% from final yr.
This comes as NVIDIA declares plans to construct AI factories within the US and accomplice with native producers to supply NVIDIA supercomputers domestically.
The corporate launched Blackwell Extremely and Dynamo to speed up large-scale AI reasoning.
Kress additionally clarified that whereas Singapore accounted for practically 20% of Q1 billed income, over 99% of the H100, H200, and Blackwell compute income billed to Singapore got here from US-based prospects.
In the meantime, authorities in Singapore and the US are additionally investigating allegations that $390 million price of NVIDIA chip servers despatched to Malaysia had been fraudulently redirected to China.
NVIDIA can also be partnering with HUMAIN to ascertain AI factories in Saudi Arabia and unveiled Stargate UAE, a next-gen AI cluster in Abu Dhabi, with companions together with G42, OpenAI, Oracle, SoftBank, and Cisco.
Furthermore, the corporate is collaborating with Foxconn and the Taiwan authorities on an AI manufacturing facility supercomputer.
Past Chips
Huang framed the difficulty past particular person chip fashions or product strains. “The AI race isn’t just about chips. It’s about which stack the world runs on. As that stack grows to incorporate 6G and quantum, U.S. world infrastructure management is at stake.”
He emphasised that “export controls ought to strengthen US platforms, not drive half of the world’s AI expertise to rivals.”
Pointing to China’s DeepSeek and Qwen fashions, Huang mentioned, “DeepSeek-R1, like ChatGPT, launched reasoning AI that produces higher solutions the longer it thinks. Reasoning AI allows step-by-step drawback fixing, planning and gear use, turning fashions into clever brokers.”
All mentioned and carried out, Huang maintained that US platforms should stay the popular platform for open-source AI even when it means collaborating with high builders globally, together with in China.”
The query stays whether or not China may have its personal AI capabilities and chip dominance or run on the American tech stack.
The publish The $50 Billion Market That’s Haunting NVIDIA appeared first on Analytics India Journal.
