Workflow
Nova 2 models
icon
Search documents
Amazon appoints longtime AWS exec Peter DeSantis to lead new AI org
TechCrunch· 2025-12-17 18:06
Core Insights - Amazon is establishing a new AI-focused organization led by Peter DeSantis, a long-time AWS executive, to enhance its AI capabilities and development [1][2] - The new organization will oversee Amazon's AI models, including Nova, as well as advancements in silicon development and quantum computing [1][3] - Amazon's commitment to AI is underscored by significant investments, including a $50 billion investment in U.S. AI infrastructure and ongoing discussions to invest $10 billion in OpenAI [4] Group 1 - Peter DeSantis has been with Amazon for 27 years, including eight years as Senior Vice President for AWS, which powers about one-third of the internet [2] - The recent AWS re:Invent event highlighted Amazon's dedication to AI for enterprise applications, justifying the formation of a new team from AWS leadership [2] - The launch of Nova 2 models and the rapid growth of custom silicon are key factors prompting the creation of this new organization [3] Group 2 - Amazon's strategy to strengthen its position in the AI sector appears to focus more on investments rather than solely on internal innovations [4] - The company has already invested $8 billion in Anthropic, a competitor to OpenAI, indicating a competitive approach in the AI landscape [4]
Amazon Introduces Four New Frontier Nova Models, a Pioneering Nova Forge Service for Organizations to Build Their Own Models, and Nova Act for Building Reliable Browser Agents
Businesswire· 2025-12-02 18:30
Core Insights - Amazon is expanding its Nova portfolio with four new models and introducing an "open training" service for organizations to create custom model variants [1] - The Nova 2 models are noted for their industry-leading price-performance in various applications including reasoning, multimodal processing, conversational AI, code generation, and agentic tasks [1] - Nova Forge is a new service that allows companies to build their own operational models [1]