Microsoft has set a target to acquire 1.8 million AI chips by the end of 2024, as per a document obtained by Business Insider. The company aims to enhance its generative AI capabilities in terms of speed, quality, and cost with a heavy reliance on purchasing GPUs primarily from Nvidia. Microsoft plans to triple its GPU count by 2024, demonstrating its commitment to remain at the forefront of generative AI advancements. This initiative in partnership with OpenAI comes with a significant price tag, with projections suggesting that Microsoft will spend approximately $100 billion on GPUs and data centers from the current fiscal year until 2027.
Microsoft’s representative Frank Shaw declined to comment due to the standard quiet period that public companies must observe before releasing earnings reports. However, tech giants like Meta are also amassing significant GPU stockpiles, contributing to Nvidia’s soaring valuation, which recently surpassed $2 trillion. Meta CEO Mark Zuckerberg announced plans to purchase 350,000 Nvidia H100 GPUs in 2024, aiming to acquire approximately 600,000 GPUs and equivalent chips by year-end. Nvidia’s recent unveiling of Blackwell, its latest and fastest GPU, coupled with the already powerful H100s currently available in the market (with an estimated cost of $30,000) highlights the company’s commitment to innovation.
Further insight from another document obtained by Business Insider reveals that Microsoft achieved “record-level GPU capacity” in the latter half of the previous year although precise figures were not disclosed. Analysts at DA Davidson estimate that Microsoft spent $4.5 billion on Nvidia chips in the past year. While Microsoft is working towards developing its own AI chips to reduce dependence on Nvidia, some within the company remain skeptical due to the substantial technological gap between them and progress in this domain unfolds rapidly.
