C114 News, Beijing time, April 11 evening news (Jiang Junmu) Google has expanded its long-term partnership with Intel, committing to use multiple generations of CPUs from the chip manufacturer in its artificial intelligence data centers.
According to the agreement, Intel's latest Xeon 6 processors will support AI training and inference workloads, helping the chipmaker gain a more favorable position in the AI hardware market dominated by Nvidia.
Amin Vahdat, Google's Senior Vice President and Chief Technology Officer responsible for AI infrastructure, stated that Intel's Xeon product roadmap gives Google confidence that &#34;we can continue to meet the growing demands for performance and efficiency of our workloads.&#34;
Intel CEO Lip-Bu Tan pointed out that the widespread application of AI requires more balanced systems, not just accelerators.
As this collaboration is established, the strategic importance of CPUs in AI systems is being re-emphasized— as agent-based AI workloads expand beyond GPUs, processors are becoming a bottleneck.
&#34;Modern data centers need CPUs to handle various computing tasks related to accelerators, and this becomes increasingly important as we transition from AI training to inference,&#34; analyst Jack Gold told Mobile World Live. &#34;While we often focus on GPUs and Nvidia's actions in that area, we overlook the demand for CPUs, which are responsible for handling all workloads and bear a significant amount of computing work beyond just AI acceleration.&#34;
He explained that Google realizes it needs powerful CPUs to handle many of its hosted cloud service workloads, even though the tech giant hopes to use some of its custom CPUs for less intensive tasks.
Gold stated that the x86 architecture still dominates the data center space, meaning Intel chips remain popular, while AMD's x86 CPUs also have strong market demand.
&#34;This is good news for Intel, which has indicated that due to strong demand, the number of CPUs they can supply to the market cannot fully meet it,&#34; he said.
Intel is producing the latest generation of Xeon processors at its factory in Arizona, USA, using advanced 18A manufacturing technology.
IPU
The two companies will also deepen their joint development of infrastructure processing units (IPUs) based on custom ASICs. Although Google is increasingly using self-developed chips, including its TPU AI accelerators, it continues to push forward.
IPUs separate network, storage, and security-related tasks from the host CPU, thereby enhancing the efficiency, resource utilization, and system predictability of hyperscale data centers.
&#34;Cloud system efficiency also depends on factors such as interconnect, storage, and power management,&#34; Gold explained. &#34;To achieve optimal efficiency, you need an IPU customized specifically for the infrastructure of hyperscale data centers.&#34;
He stated that the IPU is important for Intel as it seeks to fill the capacity of its fabs, with its manufacturing division aiming to become a large-scale chip supplier for external customers. He pointed out that the customized IPU will help Intel significantly increase the yield of high-end chips, thereby improving the company's profitability.
&#34;Ultimately, this news is good for Google as it can obtain the processing components needed to develop its cloud business. It is also good news for Intel as it maintains a relationship with a major cloud provider while keeping its factories operating at full capacity.&#34;
The financial terms and specific timeline of this collaboration have not been disclosed.
Earlier this week, Intel announced plans to participate in Elon Musk's Terafab AI chip project to jointly create processors for the billionaire's orbital data centers and humanoid robots.
<img src="https://imageproxy.pbkrs.com/https://www.c114.com.cn/news/images/t21.gif?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://www.c114.com.cn/news/images/t21.gif"/>

512480

04335

INTW

Google deepens cooperation with Intel, committing to use Intel's multi-generation CPUs, particularly the latest Xeon 6 processors, in AI data centers to support AI training and inference workloads. This collaboration aims to help Intel gain a more favorable position in the AI hardware market. Google executives stated that Intel's product roadmap will meet the growing demands for performance and efficiency. Analysts pointed out that the importance of CPUs in AI systems has been re-emphasized, especially when handling computing tasks related to accelerators. Intel's x86 architecture still dominates the data center field, with strong demand

- Google has expanded its partnership with Intel, committing to use multiple generations of Intel CPUs in its AI data centers.  
- The latest Intel Xeon 6 processors will support AI training and inference workloads, enhancing Intel's position in the AI hardware market.  
- This collaboration underscores the strategic importance of CPUs in AI systems, as demand for balanced computing systems grows beyond GPUs.

Breaking News! Google deepens cooperation with Intel in the field of AI chips