Over the years, Nvidia has proven its mettle in the world of AI with its powerful GPUs that have revolutionized the industry. Originally created as an experimental offshoot from graphics chips for gaming PCs, Nvidia’s AI-accelerating uber-GPUs have now become a force to be reckoned with. In the most recent completed quarter of 2023, Nvidia has sold a staggering 500,000 of its H100 AI chips, according to research conducted by Omdia. This achievement has solidified Nvidia’s position as a leader in the AI market.
With unit prices for the H100 ranging from $20,000 to $40,000, Nvidia’s revenue from these sales is estimated to be between $10 billion and $20 billion. Remarkably, Nvidia’s data center revenue for the same period stood at $14.5 billion, fittingly positioned in the middle of this range. These figures highlight the immense financial success Nvidia has achieved through the sale of its AI chips.
One intriguing aspect revealed by Omdia’s data is the identity of Nvidia’s biggest customers for its pricey AI chips. While it comes as no surprise that Microsoft is a major player in the AI realm, purchasing 150,000 H100s for a considerable cost of nearly $5 billion at the purported average unit price, what is astounding is that Facebook’s parent company, Meta, has also acquired the same number of chips. This places Meta in the same league as Microsoft, indicating the significance of AI technology in their operations. Additionally, both Google and Amazon are reported to have bought 50,000 H100s, which is three times less than what Meta and Microsoft have procured. Tesla, although lower on the list, has still invested in a notable 15,000 units, translating to approximately half a billion dollars.
Omdia predicts that Nvidia’s revenues from big data GPUs will double by 2027, signaling a promising future for the company in the AI domain. While not all of the AI chips will be dedicated to large language models and similar applications, a significant portion will be utilized in these areas. However, it is worth noting that some of Nvidia’s major customers, including Google and Amazon, are already venturing into building their own AI chips. This trend poses a challenge to Nvidia’s market dominance as these tech giants aim to become self-reliant. Additionally, competitors like AMD with its MI300 GPU and emerging startups like Tenstorrent, led by esteemed chip architect Jim Keller, are poised to provide strong competition in the AI chip market.
The astounding numbers associated with Nvidia’s AI chip sales paint a picture of its overwhelming success while underscoring the diminishing significance of the gaming graphics market. As AI continues to permeate numerous industries, Nvidia’s position as a key player is undeniable. However, as competition intensifies and other players develop their own AI chips, the landscape is expected to undergo significant changes in the years to come. The transformation brought about by Nvidia’s AI chips is irreversible, and the world must adapt to the new realities of artificial intelligence.