"Cloud Leader" Falling Behind? Amazon's "AI Strategy": "Low Cost" is the Core, No Competition for Cutting-Edge Models, Focus on In-House Chips and Custom Models

Amazon is trying to reshape its competitive position in artificial intelligence through a low-cost strategy. After falling behind competitors with its flagship AI model, the e-commerce giant and cloud computing leader appointed a new AI head and made cost-effectiveness a core competitive weapon.

According to The Wall Street Journal, Amazon’s new AI department head, Peter DeSantis, publicly stated during his first appearance: “AI has cost issues.” He believes that to truly revolutionize AI, the current cost structure must undergo fundamental changes.

DeSantis’s strategy centers on leveraging Amazon’s self-developed chips to develop generative AI models at lower costs than competitors, focusing on providing customized, high-value solutions for enterprise clients. This approach sharply contrasts with rivals like OpenAI and Anthropic, who frequently release cutting-edge models every few months with high-profile launches.

He emphasizes that Amazon’s goal is not to chase model iteration speed but to provide more cost-effective ways to meet customer AI needs while maintaining technological updates. This shift comes at a time when Amazon’s AI business is under pressure. Its flagship model Nova lags behind competitors in independent benchmarks, and its core technical team has experienced turbulence: former AI scientist Rohit Prasad left in December last year, and this week, AGI Lab head David Luan also announced his departure.

Financial pressures are mounting as well. Amazon plans to invest $200 billion this year in capital expenditures, most of which will go toward AI infrastructure, raising concerns among some investors about cash burn. The company’s stock has fallen about 14% since January.

New Leadership, Cost-Effective Strategy at the Core

Veteran tech executive Peter DeSantis took over the AI division in December. Joining Amazon in 1998, he has been with the company for nearly 28 years and is a key driver behind AWS infrastructure development, widely recognized internally for his deep technical expertise.

Amazon CEO Andy Jassy praised DeSantis for having “a track record of solving problems at the edge of technological possibility,” and his appointment is seen as a significant signal that Amazon is ramping up its foundational AI capabilities.

He highlights that Amazon’s self-developed Nova model has reached a mature stage, capable of accelerating development while reducing deployment costs. This move aims to address enterprise customers’ widespread concerns about the cost-to-benefit ratio of current AI model training and chip investments, as high computing costs are a major barrier to adopting AI services.

In-House Chips as a Cost-Reduction Key

This strategy is based on Amazon’s proprietary AI chips, Trainium and Inferentia, with the former designed for training models and the latter for inference. Amazon states that their chips are cheaper because they are tailored for specific tasks, up to 50% less expensive than comparable products from competitors. DeSantis explains:

“If we can build models on our own chips, we can do so at a fraction of the cost of pure AI model providers,”

Building on this, Amazon launched Nova Forge, allowing enterprise clients with specific needs to build customized generative AI models without paying for premium versions of ChatGPT, Claude, or Gemini. Tech advisor and senior CIO Tim Crawford notes that more companies are attracted to specialized models like Nova because they are cheaper, faster, and can be tailored for industry-specific tasks such as cybersecurity threat detection. He points out that CIOs are increasingly focused on the “result-to-price ratio,” prioritizing value for money.

The real-world example from Boston-based drug R&D firm Nimbus Therapeutics supports this logic. Its Director of Computational Chemistry, Leela Dodda, said that after testing multiple AI models, the company chose Amazon Nova partly because it was easier to train, less costly than competitors, and faster. She was particularly impressed by Nova’s low price, noting that in tests, Nova’s accuracy was comparable to a version of Anthropic’s Claude, but at only one-tenth the cost.

Massive Capital Expenditure Sparks Investor Concerns

Amazon’s planned $200 billion capital expenditure—roughly equal to the past two years’ total—has raised some investor alarms. Analysts estimate that Amazon will burn about $9 billion in cash in the first quarter alone. The stock’s decline since January reflects market doubts about whether this spending pace will translate into enough new business.

DeSantis responds by citing Amazon’s historical precedents. He notes that in Amazon’s early days, critics doubted the company’s ability to compete with large brick-and-mortar retailers; similarly, analysts were bearish on investing in AWS data centers. “I don’t think anyone would see those as bad decisions now,” he says.

It’s also notable that Amazon announced a $50 billion investment in OpenAI this Friday, indicating that while betting on its low-cost approach, it remains committed to a multi-pronged AI ecosystem. In talent competition, Amazon faces pressure as well. According to market research firm Levels.fyi, the average base salary for Amazon software engineers and research scientists is lower than Meta, OpenAI, Apple, and Anthropic, and the company has recently undergone two rounds of layoffs, with about 30,000 white-collar employees leaving. DeSantis expresses confidence in the current team and believes the company can continue attracting top talent.

Risk Warning and Disclaimer

Market risks exist; investments should be cautious. This article does not constitute personal investment advice and does not consider individual users’ specific investment goals, financial situations, or needs. Users should evaluate whether any opinions, views, or conclusions herein are suitable for their circumstances. Investment involves risk; proceed accordingly.