The rapid evolution of artificial intelligence (AI) technology has ushered in an era of unprecedented access to sophisticated tools like ChatGPT and Claude, but as companies scramble to monetise these innovations, the challenge of establishing a reliable pricing model becomes increasingly apparent. With industry giants like Microsoft, Google, and Anthropic investing billions into the development of Large Language Models (LLMs), the question of how to charge for AI services is fraught with unpredictability.
The Economics of AI Tokens
At the heart of this pricing dilemma lie tokens—the fundamental units that power LLMs and agentic AI systems. When users interact with these models, their queries are dissected into tokens, which the AI processes to generate a response. This mechanism, while effective, introduces variability; the same input can yield differing token outputs depending on numerous factors. This inconsistency complicates the task of cost estimation, leaving both companies and consumers grappling with fluctuating expenses.
Simon Gooch from Saviynt, a company focused on identity management, underscores the difficulty of projecting costs over extended periods. “Trying to tie someone into a cost model for the next 12 months, two years, three years, it doesn’t make any sense, honestly, because we don’t know,” he stated. The unpredictable nature of token consumption poses a significant challenge for organisations looking to budget for AI usage.
Surging Demand Amidst Rising Costs
Despite recent reductions in the price of individual tokens, as reported by Goldman Sachs, the volume of tokens consumed is expected to soar dramatically. The bank forecasts a staggering 24-fold increase in external token consumption by 2030, reaching 120 quadrillion tokens monthly. This surge corresponds with a broader shift towards integrating AI agents into business operations, further complicating cost management.
Businesses often find themselves unaware of their token utilisation until they encounter unexpected bills or depletion of their token allowances. Even industry leaders like Microsoft have had to reassess their usage of third-party coding tools due to rapid token consumption. As Will Venters, an associate professor at the London School of Economics, observes, “People are finding it really hard to manage that cost… it’s a non-deterministic output, so it’s a non-deterministic value.”
Strategies for Managing AI Costs
Organisations are beginning to adapt their strategies to mitigate the financial impact of AI. Smaller companies, for example, may exploit flat-fee personal accounts to avoid the scrutiny of larger vendors. Oliver King-Smith, founder of smartR AI, notes that while this approach may work temporarily, it is unsustainable in the long run as larger companies seek to tighten control over their pricing structures.
Rob Steele, CFO at UK accounting software firm iplicit, advocates for precision in prompting AI systems. “You wouldn’t send someone in your family out to get the weekly shop without detailed instructions,” he explains, drawing a parallel to the need for clear and specific requests when using AI to reduce token waste.
However, as firms scale their AI applications—especially when deploying them across teams or departments—costs can escalate rapidly. Venters highlights the potential for unforeseen token expenses in areas such as testing, security, and compliance, which may not have been initially considered in budgeting.
The Pricing Conundrum for AI Services
As AI companies grapple with establishing viable pricing strategies, the conversations within the industry remain fluid and complex. Bill Peterson, senior director of product marketing at Sumo Logic, describes ongoing discussions about potential pricing models for their new security services that leverage agentic AI. “We’re still having some fun conversations about this internally,” he quips, illustrating the uncertainty that permeates the sector.
Possible pricing strategies range from blanket increases to performance-based fees or bundled services. However, any pricing structure adopted may be rendered obsolete by shifts in the pricing policies of LLM providers, creating a cycle of unpredictability that few can navigate effectively.
Why it Matters
The challenges of pricing AI services underscore a critical juncture for the technology sector. As firms strive to balance innovation with profitability, the implications for consumers and businesses alike are significant. An inability to establish clear, consistent pricing models could stifle adoption, deter investment, and ultimately shape the trajectory of AI development. As the industry evolves, stakeholders must engage in open dialogue, ensuring that the benefits of AI technology are accessible while also fostering a sustainable economic model that can adapt to the complexities of this rapidly changing landscape.