A market value of $4 trillion is just the starting point! NVIDIA's overseas commercialization will ignite a new round of buying opportunities

The Zhitong Finance APP noted that NVIDIA (NVDA.US) has set a record in business history with a market capitalization of $4.17 trillion, becoming the most valuable company in the world today. Considering that NVIDIA's market value was only about $360 billion before the AI boom fully erupted in 2020, even the most pessimistic short sellers would be shocked by this growth.

Analyst Rick Orford believes that it is not too late to invest in NVIDIA. He reiterated his "buy" rating on NVIDIA, not because of its $4 trillion market value or all the favorable factors that have been analyzed, but because NVIDIA now has a brand new chip sales market.

In addition, investors should pay attention to the following key drivers. Despite the astonishing growth, NVIDIA's overseas commercialization has just begun, and the ceiling has not yet been reached.

Resumption of H20 GPU Sales to China

Previously, export restrictions caused NVIDIA to lose $4.5 billion on H20 chips, and the next quarterly report is expected to reflect an $8 billion revenue gap. However, the situation has turned around.

The U.S. government has promised to issue import licenses, and NVIDIA is about to resume supplying H20 and the new RTX PRO GPUs to China. Although the performance is not as good as the previous generation, the new products equipped with the Blackwell architecture will still maintain a premium. Chinese AI companies will undoubtedly rush to purchase, and these transactions will significantly boost the company's revenue.

It is certain that Chinese hyperscale data center operators, long constrained by hardware embargoes, will not miss this opportunity. But the business opportunities go far beyond this.

China Embraces the AI Open Source Ecosystem

After the open-source model Deepseek emerged at the end of 2023, tech giants like Baidu and Alibaba have joined the open-source wave one after another. Today, Jensen Huang praised China's open-source AI strategy as "a catalyst for global progress" while attending the China International Supply Chain Expo in Beijing. Even with the resumption of sales to China, NVIDIA's market size will still expand exponentially by lowering the barriers to AI development.

It is expected that Chinese companies similar to CoreWeave will launch affordable cloud GPU services based on NVIDIA's H20 chips. All developers will eventually converge on NVIDIA's hardware platform—after all, in the field of AI chips, both American competitors and local manufacturers find it difficult to match NVIDIA's expertise.

The Moat of the CUDA Ecosystem

Even ignoring the hardware performance advantages, NVIDIA's CUDA platform is still far ahead. Most open-source AI models have been optimized for CUDA, which means that even if new competitive architectures emerge in China, their training and tuning processes will still default to relying on NVIDIA's infrastructure.

The timing of the open-source wave is perfect for NVIDIA. The H20 is the strongest GPU that can be legally sold to China, naturally enjoying pricing power. This trend will also help NVIDIA navigate the geopolitical fog—the core "national security" reasons for export restrictions will gradually dissipate with the opening of technology.

Orford believes that the previous U.S. government ban may have been a bargaining chip in rare earth negotiations, and this game has clearly reached a compromise. Although the specific shipment volume to China has not yet been disclosed, the explosive growth of the Chinese data center market is already foreseeable. Even more intriguing is the possibility that if other countries follow the open-source model, NVIDIA's market value may break through the boundaries of imagination

Saudi AI Factory Cooperation Plan (HUMAIN)

The cooperation with Saudi HUMAIN is another important catalyst. This kingdom, which has accumulated wealth from oil for decades, is spending billions of dollars to build AI factories equipped with "hundreds of thousands of GPUs," aiming to become an AI superpower. The first phase of the project includes 18,000 NVIDIA GB300 Grace Blackwell supercomputers.

It is important to note that the demand for computing power in AI evolution is endless, which means that more NVIDIA chips will be needed in the future. This cooperation alone opens up a huge runway for the company.

UAE Stargate Project

The "Stargate" project across the Persian Gulf is equally impressive. This 1GW computing cluster, covering an area of 10 square miles (about the size of Manhattan's financial district), will use NVIDIA GB300 processors. The first phase of the 200MW cluster is expected to be operational by 2026, requiring 100,000 NVIDIA chips.

UAE's G42 is responsible for infrastructure, while OpenAI and Oracle are in charge of operations. As the demand for computing power grows, NVIDIA will continue to supply chips—this means huge profits.

Quantum Computing Layout

To maintain its lead, NVIDIA is entering the field of quantum computing. Japan's ABCI-Q supercomputer is equipped with over 2,000 H100 GPUs, which may pave the way for "the largest computing revolution since the invention of semiconductors."

Although it does not directly produce quantum processors, NVIDIA is positioning its GPUs as the bridge hardware between classical computing and quantum computing, replicating its successful model in AI infrastructure. As Orford metaphorically described during the AI gold rush: NVIDIA has always been the one selling "pickaxes." If the judgment is correct, this scene will reappear in the field of quantum computing.

Its NVAQC research center is developing GPU hardware to support quantum supercomputing. If successful, NVIDIA's stock price may experience a "quantum-level" leap.

NVLink Integration Strategy

The most underrated progress is the NVLink open strategy. On the surface, it seems that NVIDIA has given up its proprietary technology advantage, but in reality, it has built a more sophisticated ecological trap—large-scale customers will be locked into the NVIDIA system when customizing chips.

Companies like Google, Amazon, and Meta originally aimed to reduce dependence by developing their own AI chips, but NVIDIA ensures that their ASIC chips must integrate with NVIDIA infrastructure through NVLink, thereby securing product demand. This is similar to Microsoft selling Office software on Mac computers—not only does it not harm the Windows system, but it also strengthens cross-platform dominance.

As companies like Marvell, Qualcomm, and Fujitsu sign contracts, it is evident that NVIDIA has opened up new revenue streams with almost zero R&D costs.

Wall Street Consensus

In the past three months, 68 analysts have given ratings: 49 "strong buy," 12 "buy," and only 6 "hold" recommendations. The only bear gave a "sell" rating. Although NVIDIA's stock price is at a historical high, this is by no means a reason to sell at present

Conclusion

Long-term tracking has convinced Orford that there are currently no candidates that can replace NVIDIA. Just as streaming must mention Netflix, AI chips must belong to NVIDIA. This irreplaceability explains why countries like Saudi Arabia and the UAE are willing to pay a premium. With the easing of trade issues with China, NVIDIA not only gains a more certain development environment but also welcomes a geometrically growing customer base