NVIDIA's annual R&D expenses exceed $12 billion, twice that of AMD
Research and development (R&D) expenses are one of the highest quality indicators of future revenue growth. Given NVIDIA's increasing expectations for revenue in the coming quarters, it is not surprising that in terms of research and development metrics, NVIDIA is definitely ahead of its peers, including AMD.
As of the end of July quarter, Nvidia's R&D expenditure was $3.09 billion. Calculated on an annual basis, this GPU manufacturer's cumulative (possibly) R&D target for the previous quarter and the next three quarters is $12.36 billion.
In contrast, AMD spent $1.593 billion on research and development expenses in the quarter ending at the end of June. If this number is calculated annually, the company's cumulative (possibly) R&D target for the previous quarter and the next three quarters would be $6.372 billion.
In other words, according to current annualized forecasts, Nvidia's annual R&D investment is approximately twice that of AMD. This is vastly different from the situation in 2013, when the two companies invested almost identical amounts in their respective research and development activities. However, since then, Nvidia has continuously increased its investment in research and development, and is now enjoying the fruits of these investments.
More importantly, AMD had to allocate research and development resources between its CPU, GPU, and FPGA departments. In contrast, Nvidia is dedicating more and more resources to building artificial intelligence (AI) racks, which will only further solidify the GPU manufacturer's leading position in this fiercely competitive field.
William Blair believes that Nvidia's revenue for the fiscal year 2025 will reach as high as $110 billion. And the company's revenue for the 2023 fiscal year is only $27 billion.
It should be noted that the Nvidia GB200 CPU+GPU combination is currently priced between $60000 to $70000, while the price of a single Nvidia B100 AI accelerator ranges from $30000 to $35000. More importantly, the company's 72 chip AI racks are currently priced between $2 million and $3 million.
Nvidia CEO Huang Renxun recently asserted in an interview that in the next 4 to 5 years, existing data centers will require GPUs worth approximately $1 trillion to adapt to the constantly changing demand environment and achieve modernization.