Summary:
1. Google Cloud launches Vertex AI Training, a service that allows enterprises to build and train their own AI models using managed Slurm environment, data science tooling, and large-scale model training chips.
2. The service aims to attract more companies to Google Cloud by offering customization options for AI models, targeting organizations planning large-scale model training.
3. Model training can be expensive and challenging, but services like Vertex AI Training provide a solution by offering access to a managed environment for efficient and reliable training.
Article:
Google Cloud has introduced a new service called Vertex AI Training, designed to cater to enterprises looking to build and train their own AI models. This service provides access to a managed Slurm environment, data science tooling, and large-scale model training chips, making it easier for companies to customize their AI models according to their specific needs. By offering customization options and targeting organizations planning large-scale model training, Google Cloud aims to attract more enterprises to its platform.
Model training can be a costly and complex process, especially when organizations have to compete for GPU space. Hyperscalers like AWS and Microsoft have emphasized the value of their massive data centers and high-end chips, which provide enterprises with not only access to expensive GPUs but also full-stack services to support production. However, services like Vertex AI Training differentiate themselves by offering a managed Slurm environment that handles job scheduling and automatic recovery of failed jobs, resulting in higher throughput and more efficient training for a larger scale of compute clusters.
While the option to build and customize AI models may seem enticing, it may not be suitable for every enterprise. Nonetheless, services like Vertex AI Training make it easier for organizations to create niche models or completely customize existing models. This service has already attracted early customers such as AI Singapore and Salesforce’s AI research team, showcasing its potential to revolutionize the model training process for enterprises of varying sizes.