Collapse in the price of AI tokens and increasing regulation threaten billions invested

The pricing power of artificial intelligence is showing signs of fragility. Silicon's Token Spending Index LLM Data recorded a drop of almost 20% since May, while governments around the world accelerate the creation of regulations for the sector. This scenario raises doubts about the return on massive investments, over $700 billion in AI infrastructure, as reported by the Los Angeles Times.

What does the fall in the Token Spending Index mean for the pricing of AI?

Tokens are processing units used by language models (LLMs) to generate text, images or analysis. Silicon Data index measures average spending per token, and its fall indicates that the cost per unit of use of AI is decreasing rapidly. For companies that rent AI capacity, this seems positive — but the counterpart is the pressure on the margins of infrastructure providers, who need to sell much more volume to maintain revenue.

Why does pricing AI tokens matter to Brazilian investors and companies?

In Brazil, startups that use LLM APIs to create AI products — such as chatbots, translation or data analysis — already feel the impact. With cheaper tokens, the operating cost drops, enabling applications previously unfeasible. However, Brazilian venture capital investors, who closely follow global trends, begin to question whether AI infrastructure companies will be able to monetize enough to justify billionaire valuations. One positive note is that NVIDIA released scale AI computing for multi-inquiry plants, which can help dilute fixed costs.

How does regulation impact the pricing power of AI tokens?

Increased regulation — especially in the European Union and the United States — adds layers of compliance and audit that make the operation more complex. Laws such as the EU AI Act require detailed documentation, bias testing and algorithmic transparency, costs that can often not be fully passed on to final customers. This further compresses the margins of the providers and reduces their ability to tighten aggressively.

What to expect for the future of the AI infrastructure with low pricing?

With over $700 billion already committed to specialized data centers and chips for training and inference, the risk of overcapacity is real. If demand does not grow at the same speed as supply — and the price drop suggests that supply is overcoming demand — companies such as Microsoft, Google and Amazon may face difficulties to justify investments to shareholders. The consolidation of the sector and the search for differentiation in vertical applications should intensify.

Frequently Asked Questions about pricing of tokens AI

Silicon token spending index Is Data reliable?

Yes, Silicon Data is a recognized consultancy in the semiconductor and artificial intelligence sector, and its index is widely used by market analysts.

How does the fall in token prices affect the final consumer?

End-users can benefit from cheaper AI services, but in the long term pressure on providers can lead to quality reduction or increased usage restrictions.

Will regulation kill innovation in AI?

Not necessarily. Well designed settings can create a reliable environment, but the cost of compliance can slow down experimentation and favor large players over startups.