In a bold departure from Silicon Valley's conventional wisdom, artificial intelligence firm Anthropic is charting a course to compete with giants like OpenAI and Google by prioritising efficiency and disciplined spending over massive capital infusion. The company's president and co-founder, Daniela Amodei, outlined this 'do more with less' philosophy, arguing it has consistently delivered top-tier AI models despite having fewer resources.
The Frugal Frontier: Anthropic's Counter-Strategy
While the dominant narrative in the AI arms race has centred on securing record-breaking funding, locking in long-term chip supplies, and constructing sprawling global data centres, Anthropic is betting on a different formula. Daniela Amodei, in a recent CNBC interview from CES 2025, stated that Anthropic has operated with "a fraction" of the compute and capital available to its rivals. Yet, she claims, the startup's Claude models have remained among the most powerful and performant for several years.
"We have always aimed to be as judicious with the resources that we have while still operating in this space where it's just a lot of compute," Amodei explained. This approach stands in stark contrast to firms like OpenAI, which has reportedly committed a staggering $1.4 trillion towards next-generation chips, data centre campuses, and infrastructure.
Efficiency Over Exponentials: Rethinking Model Development
Anthropic's strategy involves a sharp focus on algorithmic efficiency, smarter model deployment, and leveraging higher-quality training data. The company is moving beyond the pure 'scaling laws'—the belief that simply making models bigger with more data and compute is the only path forward—pioneered by its own co-founders.
Instead, Anthropic is leaning into advanced post-training techniques and product choices designed to make its AI models cheaper to run and easier for businesses to adopt. Much of Anthropic's revenue stems from its enterprise-first model, where companies pay to integrate Claude, Sonnet, and Opus models into their workflows. Tools like Claude Code, launched in June 2025, have quickly gained popularity as command-line programming aids.
Amodei acknowledged the surprising persistence of exponential growth in AI capabilities but highlighted a crucial distinction. "There is an important distinction between the technology curve and economic curve," she said. The real challenge, according to her, is how quickly businesses and individuals can actually leverage the advancing technology.
Global Context and the Compute Question
Anthropic's frugal philosophy finds an echo in other parts of the world. Last year, Chinese AI startup DeepSeek triggered market discussions by claiming its open-weight model performed on par with cutting-edge models while using fewer resources. This led to debates about future demand for advanced chips like Nvidia's GPUs, with Microsoft CEO Satya Nadella invoking "Jevons paradox" on social media platform X, suggesting efficiency could paradoxically increase overall demand.
Despite its efficient ethos, Anthropic does not deny the growing hunger for computing power. Amodei admitted that future compute requirements are "very large" and that the company will need more to "stay at the frontier." However, the core of its strategy remains achieving superior results through smarter resource allocation rather than merely outspending competitors.
This calculated approach by Anthropic, led by siblings Daniela and CEO Dario Amodei, presents a compelling alternative narrative in the high-stakes AI industry. It questions whether the future of AI leadership will be won by the biggest bank account or by the most ingenious application of available tools.
