Amazon just dropped a premium AI service that could reshape how enterprises build custom models. Nova Forge lets AWS clients customize AI models during training for $100,000 per year - a fraction of the hundreds of millions it typically costs to build from scratch. The service represents Amazon's bold bet on capturing enterprise AI spending as it trails behind Anthropic and OpenAI in market share.
Amazon is making a major play for enterprise AI dollars with Nova Forge, a premium service that lets companies customize AI models during the training process rather than after completion. The $100,000 annual price tag might sound steep, but it's a bargain compared to the hundreds of millions or even billions companies typically spend building large language models from scratch.
The timing couldn't be more critical for Amazon's AI ambitions. According to Menlo Ventures data, Amazon's Nova models captured less than 5% of the enterprise AI market by mid-2024, while Amazon-backed Anthropic leads with 32%, followed by OpenAI at 25% and Google at 20%. Meta even claims 9% - nearly double Amazon's share.
"We are a frontier lab that has focused on customers," Rohit Prasad, Amazon's head scientist for artificial general intelligence, told CNBC. "Our customers wanted it. We have invented on their behalf to make this happen."
The key differentiator with Nova Forge is timing. While most AI customization happens after a model finishes training through fine-tuning, Forge gives clients access to models at various stages of development. This means customer data gets baked in earlier, creating models that lean more heavily on proprietary information rather than generic training data.
Reddit became an early showcase for the technology's potential. The social platform needed an AI model sophisticated enough to moderate content across countless specialized communities. Engineers found that a Nova model enhanced with Reddit data through Forge outperformed commercially available large-scale models, according to Prasad. Other early adopters include Booking.com, Nimbus Therapeutics, the Nomura Research Institute, and Sony.
The service addresses a genuine pain point in enterprise AI adoption. Most companies want AI that understands their specific domain, customer base, or technical requirements, but building custom models requires massive computational resources and AI expertise that few organizations possess. Nova Forge essentially lets them rent Amazon's infrastructure and expertise while maintaining control over their proprietary data integration.
