Google has announced the launch of its custom-built Arm-based server chip, Axion, set to hit the market in 2024. The revelation came during Google's Cloud Next conference in Las Vegas, marking a significant step for the tech giant in catching up with its competitors, Amazon and Microsoft, who have already embraced similar strategies to fortify their cloud infrastructure offerings.
This strategic pivot underscores Google's intent to carve a more substantial niche in the cloud services market, an area where it has traditionally lagged behind Amazon Web Services (AWS) and Microsoft Azure. With cloud computing burgeoning, Google's foray into custom chip design with Axion aims to provide a more cost-effective and efficient alternative to the dominant x86 server models, primarily powered by AMD and Intel processors.
Axion's introduction aligns with Google's broader efforts to diversify its revenue streams beyond its advertising stronghold. Despite advertising accounting for the lion's share of Alphabet's revenue, Google's cloud segment has shown promising growth, representing a significant portion of the company's overall revenue. This shift towards cloud computing is not just a strategic business move but also part of Google's commitment to sustainability, with Axion expected to offer a 60% boost in energy efficiency compared to traditional x86-based virtual machines (VMs).
Google's venture into Arm-based server technology isn't entirely new. The company previously provided access to virtual machines utilizing Arm-based chips from Ampere, an Oracle-backed startup. However, Axion represents Google's first in-house developed Arm CPU, signaling a deeper commitment to this technology.
The move to Arm architecture is partly driven by economic considerations, as organizations look to trim cloud computing expenses amidst broader financial concerns. Arm's architecture, known for its efficiency and lower power consumption, has been highlighted as a cost-effective alternative, with potential savings in operational expenditures for cloud services.
The performance metrics of Axion are compelling, with Google touting a 30% performance improvement over the fastest general-purpose Arm-based VMs in the cloud and a 50% edge over x86-based counterparts. This performance enhancement is crucial for data-intensive and compute-heavy applications, particularly those leveraging artificial intelligence and machine learning technologies.
Google's TPU v5p, another innovative chip designed for artificial intelligence workloads, further cements the company's position in the high-performance computing space. The TPU v5p, now generally available through Google Cloud, is engineered for optimal performance and efficiency, supported by advanced liquid cooling systems to manage the thermal output of its massive 8,960-chip pods.
