Building up an LLM (Large Language Model) on your own, as advised by OpenAI Chairman Bret Taylor, could lead to massive financial depletion.

Building up an LLM (Large Language Model) on your own, as advised by OpenAI Chairman Bret Taylor, could lead to massive financial depletion.

In a recent podcast interview, OpenAI chairman Bret Taylor warned that building proprietary large language models (LLMs) is a financially risky and capital-intensive endeavour for businesses, costing hundreds of millions – even billions – of dollars.

The AI industry is undergoing consolidation, with a few well-funded giants like OpenAI, Google, Meta, and Anthropic dominating the space. This trend is driven by the astronomical costs associated with training state-of-the-art LLMs. For instance, the costs can reach upwards of $100 million, with some projections reaching the billion-dollar mark.

Taylor's warning reflects this consolidation trend. He advises businesses to leverage existing AI platforms and APIs instead, which provide access to advanced AI capabilities without the massive financial and operational burdens of developing and training their own LLMs. Using APIs allows businesses to incorporate advanced AI capabilities more efficiently by paying for usage rather than investing billions upfront.

OpenAI itself has adopted this strategy in 2025, offering advanced AI models through platforms and partnerships, rather than focusing on enterprise-level development. This approach may accelerate the development of more specialized AI models tailored to specific industries or use cases.

The focus should be on identifying the most suitable AI platforms and APIs to integrate into products and services. C-suite executives should reevaluate their company's AI strategy due to the astronomical capital requirements for building proprietary LLMs. Partnerships with established players like OpenAI or Google, or exploring niche providers that cater to specific industries or use cases, may be necessary.

This shift in strategy will likely lead to increased demand for AI-as-a-Service offerings from companies like OpenAI, Google, and Microsoft Azure. However, the consolidation of AI capabilities raises concerns about market competition, innovation, and the potential for vendor lock-in. Businesses across sectors will need to carefully evaluate their AI strategies, balancing the benefits of leveraging existing platforms with the risks of over-reliance on a limited number of providers.

In conclusion, the high costs of building proprietary LLMs have made it a non-viable option for most businesses. Instead, businesses are likely to rely on existing AI platforms and APIs to integrate cutting-edge language capabilities into their products and services. This may lead to increased merger and acquisition activity, as smaller AI companies struggle to keep pace with the capital requirements. The AI landscape will continue to evolve rapidly, with the gulf between the dominant players and the rest of the market likely to widen as the costs of training frontier LLMs continue to rise.

1. Building proprietary large language models (LLMs) is a financially risky and capital-intensive endeavor for businesses, costing hundreds of millions – even billions – of dollars, as highlighted by OpenAI chairman Bret Taylor.
2. Due to the astronomical costs associated with training state-of-the-art LLMs, the AI industry is undergoing consolidation, with a few well-funded giants like OpenAI, Google, Meta, and Anthropic dominating the space.
3. Taylor advises businesses to leverage existing AI platforms and APIs instead, which provide access to advanced AI capabilities without the massive financial and operational burdens of developing and training their own LLMs.
4. In 2025, OpenAI itself adopted this strategy, offering advanced AI models through platforms and partnerships, rather than focusing on enterprise-level development.
5. C-suite executives should reevaluate their company's AI strategy due to the high capital requirements for building proprietary LLMs, with partnerships with established players like OpenAI or Google, or exploring niche providers that cater to specific industries or use cases, being a viable option.
6. This shift in strategy will lead to increased demand for AI-as-a-Service offerings from companies like OpenAI, Google, and Microsoft Azure, but raises concerns about market competition, innovation, and the potential for vendor lock-in.

Read also: