Google has also launched the A3 Mega instance for developers needing more power for large language model training, and the A3 confidential instance for secure data and AI workload processing. The company also launched its most powerful homegrown AI accelerators, the Cloud TPU v5p processors, and announced new AI-optimized storage options. Google Cloud is also launching more traditional instances powered by Intel's fourth- and fifth-generation Xeon processors, including the new general-purpose C4 and N4 instances, the C3 bare-metal machines, and the Z3, a storage-optimized virtual machine.
Key takeaways:
- Google Cloud Next announced new instance types and accelerators for the Google Cloud Platform, including AI accelerators built by Google and Nvidia.
- Support for Nvidia's high-performance Nvidia HGX B200 for AI and HPC workloads and GB200 NBL72 for large language model training will arrive in early 2025, with the GB200 servers being liquid-cooled.
- Google launched its Cloud TPU v5p processors, the most powerful of its homegrown AI accelerators yet, and announced new AI-optimized storage options, including Hyperdisk ML.
- Google Cloud is also launching a number of more traditional instances, powered by Intel’s fourth- and fifth-generation Xeon processors, including the new general-purpose C4 and N4 instances, and the Z3, Google Cloud’s first storage-optimized virtual machine.