GPU factory NVIDIA (Nvidia) announced the second quarter financial report of 2026 on the 28th. The total customer contribution of the two major secrets was 39%. The highly concentrated customer structure has attracted the outside world to pay attention to the stability of NVIDIA's business.
According to CNBC, NVIDIA's second quarter financial report "Customer A" accounted for 23% of the total revenue, and "Customer B" accounted for 16%, which was a clear comparison with the same period of financial year 2025. At that time, the total revenue of NVIDIA's top two customers accounted for 14% and 11%, indicating that the trend of customer concentration increased. NVIDIA also admits that the majority of the revenue is earned from a small number of customers, and this trend may continue.
Although NVIDIA has not disclosed the exact identity of the two mysterious customers, it is widely recommended that these customers may be large cloud service providers such as Microsoft, Amazon, Google and Oracle, and NVIDIA only calls them "direct customers." NVIDIA's explanation of "direct customers" is not the final user of chips, which are companies that buy NVIDIA chips, integrate them into complete systems or circuit boards, and then transfer them to data centers, cloud suppliers or end users. These customers may also include original design manufacturers (ODMs) or original equipment manufacturers (OEMs) such as Foxconn or Quanta, or sellers or system integrators such as Dell.
"Intercom customer pickup" is usually a cloud service provider, network company and enterprise that buys systems from NVIDIA's direct customers. It is worth noting that NVIDIA also revealed that two "client pickup customers" paid more than 10% of their total revenue through customer A and customer B. In addition, an AI research and development company has also brought NVIDIA an affordable acquisition through direct and intermediary customer access channels. Due to this flexibility in customer definition, it is more challenging to distinguish the real identities of these mysterious customers.
Despite the customer concentration, the market is in trouble. However, NVIDIA still stated at the financial conference that the market's demand for its AI systems is still strong. The source of demand is not only from traditional cloud suppliers, but also companies that buy systems for AI, new suppliers focused on AI services, and foreign governments. NVIDIA financial manager Colette Kress pointed out that large cloud service providers account for about 50% of the company's data center revenue, while the overall data center business itself accounted for 88% of the total NVIDIA's revenue in the second quarter, highlighting its key position in the company's overall revenue.
At present, market analysts are paying close attention to the capital expenditure status of cloud service providers to predict the future growth of NVIDIA. Analysts pointed out that if there is no clearer prospect of capital expenditure expectations for cloud service providers, NVIDIA's recent upward profit revision or stock price increase will be limited. However, NVIDIA executive chief Huang Rensheng is confident in the AI market prospects. He further predicted that by the end of this century, the total investment in AI-based facilities will still reach $3 to $4 trillion.
Huang Rensheng emphasized that in a $50 billion AI data center, NVIDIA not only provides its GPU, but also includes other chips, which may account for about 70% of the total cost. He stressed that capital spending for large cloud service providers has doubled in the past two years, and is expected to invest $60 billion in 2025. In addition, new customers such as enterprises and overseas cloud suppliers are also investing heavily in the construction of AI-based infrastructure, which makes NVIDIA even more expected to receive approximately US$20 billion in revenue from the dominant AI field in 2025.