AI Model Commoditisation Risk

Star investor Mary Meeker warns that OpenAI faces increasing competition from rivals like China's DeepSeek, potentially leading to the commoditisation of large AI models. Meeker suggests this competition could create numerous companies valued at trillions of dollars, not necessarily based in North America. She highlights the rising costs of training models like OpenAI's GPT series and Google's Gemini, while cheaper models from companies such as DeepSeek put pressure on pricing.

Meeker's analysis points to a shift in the AI business model, questioning the one-size-fits-all approach of large language models. She notes the emergence of smaller, more specialised models trained for specific uses. DeepSeek's R1 model, for example, rivals leading AI systems at a fraction of the cost. This affordability and performance have drawn international interest, shaking OpenAI and prompting a re-evaluation of its strategies, including a potential move towards more open-source approaches. DeepSeek's advancements have disrupted the industry, challenging the dominance of US companies and pushing OpenAI to accelerate its development pipeline and reaffirm its commitment to substantial infrastructure investment.

DeepSeek's competitive pressure has also led OpenAI to mimic certain strategies, such as showcasing AI model reasoning and offering a mini reasoning model for free. This competitive landscape signals a transition in the AI market, with implications for business models and the distribution of AI development and wealth creation.