The rising cost of using large language models (LLMs) is becoming a critical issue for enterprises looking to integrate artificial intelligence into their operations, according to Palo Alto Networks CEO Ashish Arora. Speaking on CNBC’s “Squawk on the Street,” Arora emphasized that token costs—essentially the price paid per unit of text processed by AI systems—are slowing down widespread adoption of AI technologies.

The Cost Challenge for Enterprises

Token costs have become a central concern for businesses as they scale their use of AI. These costs are measured in dollars per token, where one token is roughly equivalent to a word or subword in the input text. As companies deploy more AI tools and models, especially in areas like customer service, data analysis, and software development, the financial burden has grown significantly.

Arora’s comments reflect a growing sentiment among enterprise leaders who are struggling with the economics of AI deployment. He stated that token costs need to drop by 20% over the next year and by 90% within the following year for widespread adoption to take place. This is not just about cost savings—it’s about making AI accessible, scalable, and sustainable for businesses of all sizes.

A Shift in AI Usage Strategies

In response to rising costs, many companies are rethinking how they use AI tools. Some have introduced usage caps to limit the amount of text processed by their models each month. Others are encouraging employees to adopt more efficient practices, such as using the right tool for the right task and sharing cost-saving strategies across teams.

One notable trend is the shift toward older, cheaper models. While cutting-edge large language models like GPT-4 offer advanced capabilities, they come at a high price. Companies are increasingly looking at alternatives that provide sufficient performance without the steep financial burden. This includes exploring open-source models, which can be customized and deployed with lower overhead.

The Role of Chinese AI Labs

The report also highlights an opportunity for Chinese AI labs, which are able to offer more cost-effective solutions due to their efficient models and lower energy costs. While U.S.-based companies like OpenAI and Google have dominated the AI landscape, the emergence of competitive alternatives from China is creating a new dynamic in the global market.

This shift could lead to increased competition and innovation, potentially driving down token costs for all users. However, it also raises questions about data privacy, regulatory compliance, and geopolitical tensions that may affect the adoption of these models in Western markets.

The Impact of Token Shock on Silicon Valley

The financial strain caused by high token costs has already had a tangible impact on some of Silicon Valley’s largest companies. For example, Uber reportedly exhausted its full-year 2026 AI budget by April, forcing the company to reassess its strategy and consider alternative approaches.

Chief Technology Officer Praveen Neppalli Naga noted that Uber was “back to the drawing board” in terms of how it approached AI integration. Meanwhile, Chief Operating Officer Andrew Macdonald emphasized that token costs would now be weighed directly against the cost of hiring engineers—a clear indication of how financial constraints are reshaping decision-making processes.

The Financial Services Lead in Enterprise AI

According to PYMNTS Intelligence, companies in financial services and insurance, healthcare, and media and advertising are leading the charge in enterprise AI adoption. These industries have been quick to recognize the value of AI in improving efficiency, reducing costs, and enhancing customer experiences.

However, as they invest more heavily in AI, these enterprises are also becoming more selective about which projects receive real capital. The report suggests that while some initiatives are moving forward with strong backing, others are still in the proof-of-concept stage and require further validation before significant investment is made.

Conclusion

The rising cost of tokens has become a major obstacle for businesses looking to fully integrate AI into their operations. As companies like Palo Alto Networks and Uber demonstrate, the financial burden of using large language models is forcing a reevaluation of how AI is deployed and managed. With token costs expected to drop significantly in the coming years, there is hope that AI will become more accessible and affordable for a wider range of organizations.

However, until then, businesses must continue to innovate and optimize their use of AI tools. This includes exploring cost-effective models, adopting open-source solutions, and refining internal strategies to maximize efficiency without compromising performance.

As the landscape continues to evolve, readers should keep an eye on how token costs change over time and how different industries adapt to these challenges. The future of enterprise AI will depend not only on technological advancements but also on the ability of companies to manage their financial resources effectively in this rapidly changing environment.

---

Original Source

This article is based on publicly available reporting. For the complete original story, visit the publisher’s article.