NVIDIA, the world's most valuable company by market cap, recently delivered an unprecedented quarterly revenue of $81.6 billion, an 85% year-over-year surge, surpassing Wall Street expectations. However, investors' response fell short, consolidating around a "Zero-China" revenue position.
The firm released its quarterly results after its CEO, Jensen Huang, joined President Donald Trump and other tech executives in the US-China summit in May. The sale of NVIDIA chips to China, announced at the summit, was seen as a new possibility to open the local market and bring significant revenue, but so far it has faced some issues.
"In the short term, NVIDIA has already de-risked its China exposure by assuming $0 in data center compute revenue from the region," Anders Bylund, contributing tech analyst at The Motley Fool, told the International Business Times.
"Recent customs blockades on H200 and RTX 5090D V2 chips confirm that the zero-dollar assumption was reasonable. The message from Beijing to local system builders is clear: shop domestic," Bylund added.
Issues affecting the sale of NVIDIA chips include Chinese officials pushing domestic companies to use Chinese AI chips made by companies like Huawei and Cambricon, the New York Times reported.
The short and medium-term outlook for NVIDIA in China depends heavily on U.S. export policy, and "investors should not expect a U-turn," said Bylund. "The real tell will be whether Chinese AI companies like Huawei and Alibaba become the native environment for China's best AI models," he added.
Stephen Wu, former Microsoft and Amazon and current Managing Partner at Carthage Capital Management, also told International Business Times that "short-term, NVIDIA captures remaining capital by selling restricted, downgraded chips."
Mid-term, large Chinese tech companies accelerate their transition to domestic alternatives, said Wu. "Long-term, China operates a parallel compute ecosystem that completely bypasses US silicon," Wu explained.
"Investors projecting linear growth for NVIDIA in China are fundamentally mispricing the geopolitical tail risk," he added.
Despite the ongoing tech tensions between the U.S. and China, a complex near-future outlook, and a cold market response to its recent quarterly report, NVIDIA continues to signal business opportunities in China and the broader Asian region.
On June 12, NVIDIA announced a partnership with Unitree, a Chinese research-focused robotic firm. Unitree was recently approved for an IPO on the Shanghai Stock Exchange on the same day that the deal with NVIDIA was announced, CNBC reported.
Additionally, NVIDIA is going beyond chips and investing in the CPU market in Asia. During the recent earnings call, CEO Huang said that the recently debuted "NVIDIA Vera Platform" gives the company access to a new $200 billion market. When asked in Taipei if that market included China, Huang said, "I would think so."
"NVIDIA can't ship high-end training GPUs past Chinese customs, so the company is doing everything else instead," said Bylund.
The company is pouring capital into neighboring Taiwan, South Korea, and Singapore, (while) partnering with Chinese robotics firms like Unitree on products aimed at the global stage, and co-developing edge chips with Taiwanese chipmaker MediaTek for the Windows AI PC market, Bylund explained.
"The logic is simple: data center GPUs are a regulatory brick wall, so NVIDIA is climbing around it," said Bylund.
Robotics, edge AI, and consumer devices face much lighter export scrutiny and looser limits. If AI agents end up running on billions of laptops and robots rather than just hyperscale data centers, NVIDIA wants to be there, Bylund added.
The new Vera Platform, presented in the quarterly report and the successor of the Blackwell platform, was developed specifically for the Agentic AI era. Amazon Web Services (AWS), Google Cloud, Microsoft Azure, and Oracle Cloud will be the first companies to deploy Vera Rubin-based technology.
"Geopolitical headwinds mean GPUs would be heavily regulated, and exports might come under sanctions," Akshar Prabhu Desai, Software Engineer at Google, told the International Business Times.
"But high-end CPUs are generally not sanctioned, and hence it is a market NVIDIA still can expand into in other countries such as China through their Vera CPUs," said Desai.
Vera provides energy-efficient CPUs that work seamlessly with their GPUs to ensure that Agentic AI has access to both powerful CPUs and GPUs that can work together optimally, Desai explained.
Based on NVIDIA's claim, Vera can handle parallel software environments and tool calls up to 1.8x faster than traditional x86 server chips. NVIDIA is thus expanding itself from GPUs that power models into the wider AI landscape where Agentic AI is the central focus, said Desai.
"For their enterprise customers, this might mean a single-vendor unified architecture which lowers the per-token cost not just for their LLM usage but for their entire Agentic AI pipelines," said Desai.
The Vera Rubin numbers are eye-catching: 35x higher inference throughput than Blackwell, 10x better performance per watt, and NVIDIA claims it unlocks a $200 billion CPU market, said Bylund from Motley Fool.
"That's the Vera piece of the Vera Rubin puzzle, just as Grace plays the CPU role in the Grace Blackwell AI accelerator cards," Bylund explained. "NVIDIA sees the AI industry shifting from 'train giant LLM models' to 'run AI agents everywhere, all the time,' and Vera is built for that world."
What should investors be watching for? Real-world benchmarks against rival products such as AMD Instinct, Google Trillium, Intel Gaudi, and Amazon Trainium, when the hardware becomes available, Bylund told us.
"Assuming NVIDIA's performance claims are in the right ballpark, Vera Rubin should grab the lion's share of this adjacent (and massive) market," said Bylund.
"The Vera CPU platform represents a hostile takeover of the server market," said Wu.
NVIDIA is no longer just selling specialized AI chips, Wu explained. "They are using their market dominance to bundle their AI chips with their new Vera CPUs, effectively forcing buyers to purchase the entire NVIDIA system," Wu said. "This structurally locks competitors like Intel and AMD out of the server infrastructure loop."
Describing U.S. and China tech stack development, Wu said, "The endpoint is total supply chain decoupling."
"It's no longer about whether NVIDIA can ship chips past Chinese customs," said Bylund. "It's about which ecosystem becomes the default for the rest of the world."
NVIDIA will survive without China, but losing it permanently caps their maximum growth potential, said Wu. "They [NVIDIA] are scaling aggressively into global data centers, government AI projects, and physical robotics," Wu added.
From autonomous vehicles, logistics, robotics, data centers, telco, enterprise AI, and other industries, the U.S.-China tech competition is spilling globally. Companies from the E.U., Latin America, Africa, India, and South East Asia are leveling up their tech stack for the Agentic AI era, and demand is high for chips, processing power, data centers, infrastructure, edge, and platforms.
