Microsoft's AI ambitions may be facing a significant hurdle: a shortage of chips. The company's plans to install 1.8 million AI chips by the end of 2024 seem to have hit a snag, as internal documents reveal only 2.2 million chips in operation. This discrepancy raises questions about the true state of Microsoft's AI infrastructure and the potential impact on its expansion efforts.
The issue lies in the complex relationship between AI chip manufacturers, particularly Nvidia, and the tech giants that rely on their products. Nvidia's supply chain is a closely guarded secret, and the company rarely discloses sales figures or client information. This lack of transparency makes it challenging to assess the true scale of AI development across the industry.
Microsoft, a key player in this arms race, has been investing heavily in AI infrastructure. Since 2022, the company has poured over $280 billion into land, buildings, and computational resources, with a focus on AI. However, estimating the exact number of datacentres built and their operational status is a complex task. Microsoft's annual reports and quarterly earnings provide some insights, suggesting an addition of 5GW of datacentre capacity over the past two years.
This capacity, if fully utilized, would suggest a need for approximately 6.4 million graphics processing units (GPUs). However, sustainability reports indicate a lower AI capacity of around 1.2GW, implying a requirement for only 4 million AI chips. This discrepancy highlights the challenge of accurately assessing Microsoft's AI infrastructure.
The situation is further complicated by Microsoft's partnership with OpenAI, which may account for some datacentre deployments not reflected in the documents. Additionally, some of Microsoft's major projects, like the Fairwater datacentres, appear to be far from operational, with only a fraction of the planned capacity built.
Nvidia's balance sheets suggest a significant number of chip sales, but the exact distribution and installation remain unclear. Microsoft's CEO, Satya Nadella, has acknowledged the challenge of matching chip availability with datacentre infrastructure, emphasizing the need for 'warm shells' to plug into.
Calculating the number of chips from AI capacity figures is a complex task. While a broad approximation can be made by dividing power usage by AI chip power consumption, factors like cooling systems and the presence of non-AI chips on server racks must be considered. This results in a conservative estimate of around 6.4 million chips for Microsoft's AI infrastructure.
In conclusion, Microsoft's AI plans may be hindered by a shortage of chips, and the true scale of its AI infrastructure remains uncertain. The company's expansion efforts and the broader AI arms race are intricately linked to the availability and distribution of these critical components.
