The AI Arms Race Has No Moat—Just Money

Capital Is the Only Kingmaker

In a striking analysis that challenges the perceived dominance of top AI labs, veteran tech analyst Ben Thompson argues that companies like OpenAI, Google DeepMind, and Anthropic do not possess a sustainable competitive advantage—or “moat”—aside from their access to billions in capital. While these labs have enjoyed early leadership, largely due to expensive training runs and infrastructure, Thompson says their edge is inherently fragile. With foundational models increasingly commoditized, and the barrier to entry lowering due to open-source alternatives, the notion of a defensible monopoly in AI is rapidly eroding. In other words, in today’s generative AI boom, deep pockets—not deep innovation—remain the defining differentiator.

Open Source Closes In

The report highlights how open-source AI models are catching up at an astonishing pace, fueled by community contributions and accelerated research. Unlike closed models, open alternatives like Meta’s LLaMA and Mistral are gaining ground with fewer resources and greater adaptability. Thompson contends that the inability of frontier labs to build a unique moat—be it through proprietary data, distribution, or algorithms—leaves them vulnerable to disruption. This has profound implications for the future of artificial intelligence: the next AI giant could just as easily emerge from a garage as from a trillion-dollar enterprise, provided they can get their hands on enough compute.