Amazon Targets Frontier AI Models Within 1 Year as Nova2 Reaches 50,000 Customers
Updated
Updated · CNBC · Jun 17
Amazon Targets Frontier AI Models Within 1 Year as Nova2 Reaches 50,000 Customers
1 articles · Updated · CNBC · Jun 17
Summary
Peter DeSantis said Amazon expects to compete with OpenAI and Anthropic on frontier AI models within the coming year, while acknowledging its current models still lag on the biggest workloads.
Nova2, released in December, has reached about 50,000 customers as Amazon pursues a two-pronged AI strategy: selling its own models and offering third-party models through AWS Bedrock.
Trainium and Graviton chips sit at the center of that push, with DeSantis arguing Amazon should be viewed alongside Nvidia because it can design, build and deploy its own AI hardware.
AWS still mainly rents out that computing capacity—Anthropic is a major customer—but Amazon is also weighing future sales of Trainium racks and potentially Graviton chips to third parties.
The push reflects Amazon's broader effort to convince investors it can be a leading AI player after falling behind the top frontier-model labs.
Is Amazon's sold-out custom chip business a greater threat to Nvidia than its AI models are to OpenAI?
Can Amazon's platform succeed if it empowers the very AI rivals it aims to surpass?
Amazon’s Nova 2 and Nova Forge: Transforming Enterprise AI with Custom Models and Industry-Leading Performance (2025-2026)
Overview
Amazon has made a major move in AI by launching Nova Forge in late 2025, expanding its presence in the advanced AI landscape. Nova Forge is designed to help organizations with proprietary or industry-specific data build highly specialized AI models tailored to their unique needs. This service enables companies to develop AI that deeply understands their particular domains, moving beyond general-purpose models to deliver more precise and relevant insights. Nova Forge supports the creation of models that can interpret specialized information and processes, making it valuable for industries like manufacturing, automation, and research, and helping organizations unlock new levels of operational efficiency.