Nvidia has delivered its initial H200 chips to China; however, it anticipates that there will be no revenue from data centers resulting from these shipments.
Nvidia has commenced shipments of its H200 processors to China under a new US licensing framework, with these shipments representing less than 1% of the $89 billion in quarterly data center revenue. The company is forecasting $108 billion for the current quarter, assuming no revenue from data center computing in China whatsoever.
Nvidia has advised investors not to anticipate any contributions from China. The processors were sent out in the quarter ending July 26 and constituted a minimal fraction of the total revenue.
The forecast warrants a second look: Nvidia estimates $108 billion for the current quarter while presuming no data center computing earnings from China. Jensen Huang indicated that even this projection is constrained. “The unconstrained growth would be a lot higher,” he remarked, suggesting actual demand significantly exceeds the 70% growth forecast the company has provided for next year.
When viewed from this perspective, the figures tell a different story. A single supplier like Nvidia can exclude the world's second-largest economy from its forecast and still increase its guidance by nearly 20%.
In response, Europe is proposing seven facilities. The Commission has initiated bidding for up to seven AI gigafactories at a cost of €30 billion, divided roughly into €10 billion public and €20 billion private funding. However, only €1 billion of the public portion is currently secured, with the remainder contingent on the forthcoming long-term budget, which a Commission official notably refrained from preempting.
When juxtaposing these amounts, it is apparent that Nvidia generates approximately a billion dollars in data center revenue daily, meaning the committed investment for the seven gigafactories in Europe equals about one day's worth of sales for Nvidia.
Each site is designed to accommodate at least 100,000 advanced chips, all of which will be sourced on the open market from the very company that has projected $108 billion in revenue.
Unfortunately for buyers, pricing is trending upwards. Nvidia has informed its clients that server costs will increase by over 15% starting early next year. Additionally, the timeline isn’t favorable; construction is slated to begin in early 2027, with the first factories expected to be operational by mid-2028.
The Next Web has pointed out that the sovereignty narrative surrounding this issue is superficial. Acquiring the same silicon from a European source presents an illusion of autonomy rather than true independence.
Thus, the H200 narrative fundamentally revolves around leverage. Nvidia has the capacity to lose a market as large as China and still expand, while Europe is in discussions about facilities that are not yet financed and reliant on chips that it does not manufacture.
Published August 28, 2026 - 8:19 am UTC
Back to top
Other articles
Nvidia has delivered its initial H200 chips to China; however, it anticipates that there will be no revenue from data centers resulting from these shipments.
Nvidia has sent its initial H200 units to China while still predicting no revenue from data centers in that region. In response, Europe plans to invest €30 billion in gigafactories, although most of the funding remains unallocated.
