AMD CEO Lisa Su has been talking about how their artificial intelligence chips are an ideal alternative to hard-to-get NVIDIA GPUs, but two of the three major cloud computing providers do not agree.
At this week's Google Cloud Annual Technology Conference in Las Vegas, Google Cloud's Vice President and General Manager of AI Infrastructure, Mark Lohmeyer, stated that currently, Google only needs NVIDIA GPUs and their own TPU to meet AI workloads.
In an interview with the media, he said:
<blockquote>
Currently, we are deploying NVIDIA GPUs and our own TPUs.
</blockquote>
Google Cloud also uses Intel's new central processing units, which he said are &#34;suitable for certain inference workloads.&#34;
In addition to Google, Amazon AWS also does not use AMD products. A senior executive at AWS previously stated in an interview with the media that AWS mainly uses NVIDIA GPUs and Amazon's self-developed AI chips Trainium and Inferentia.
However, even without Google or Amazon, AMD remains a winner in the AI wave. According to the company's guidance, its chip sales this year are expected to reach a new high. In the latest financial report, AMD's estimated sales of AI chips this year have increased by 75% to over $3.5 billion. AMD stated that it is collaborating with Microsoft, Oracle, and Meta to install the latest GPUs in their data centers

ORCL

Google and Amazon did not use AMD's AI chips, instead opting for Nvidia and in-house chips. However, despite the lack of collaboration with these two tech giants, AMD still expects AI chip sales to reach a new high, exceeding $3.5 billion. In addition to partnerships with Microsoft, Oracle, and Meta, AMD also plans to install the latest GPUs in data centers

Google, Amazon, neither use AMD!