As artificial intelligence (AI) continues to revolutionise industries, the underlying economics of this technology—known as tokenomics—poses significant challenges for businesses and consumers alike. Major players like Microsoft, Google, and Anthropic have poured hundreds of billions into developing Large Language Models (LLMs) that power services like ChatGPT, Claude, and Gemini. While accessing these cutting-edge tools for free is a steal, the question remains: how do we fairly price AI services that are still evolving?
The Rise of AI and Its Costly Underpinnings
The demand for AI has skyrocketed, with companies increasingly utilising LLMs for a range of applications from coding assistance to automated customer service. However, the pricing model for these services is far from straightforward. Simon Gooch, a key player at Saviynt, highlights the unpredictability of pricing in this rapidly changing landscape. “Trying to tie someone into a cost model for the next 12 months, two years, or three years doesn’t make any sense, honestly, because we don’t know,” he states, a sentiment echoed across the industry.
At the heart of this issue lies the concept of tokens. In simple terms, tokens are the basic units processed by AI models when users input prompts. These prompts are decomposed into tokens that the AI interprets and then generates responses from, also in token form. The challenge? The number of tokens consumed can vary significantly based on the complexity of the request and the specific model employed, making it difficult for users to predict their costs.
The Token Consumption Explosion
Goldman Sachs predicts a staggering rise in token consumption, estimating an increase of 24 times by 2030, reaching a colossal 120 quadrillion tokens per month. This projection stems from businesses transitioning to AI-driven decision-making processes. However, both companies and individuals often find themselves unaware of how quickly they are racking up token costs—until they receive their monthly bill or run out of credits entirely.
For instance, even tech giants like Microsoft have reportedly scaled back their engineers’ use of certain third-party coding tools after noticing unexpected spikes in token consumption. Similarly, Uber faced a significant budget blowout within months of implementing AI coding solutions, highlighting just how unpredictable costs can be.
Managing the Tokenised Economy
The unpredictability of token costs poses a major headache for businesses, as expressed by Will Venters, Associate Professor of Digital Innovation at the London School of Economics. “People are finding it really hard to manage that cost… it’s a non-deterministic output, so it’s a non-deterministic value,” Venters explains. Companies must tread carefully, especially when rolling out AI systems across large teams where token consumption can escalate rapidly.
Oliver King-Smith, founder of smartR AI, notes that smaller firms often circumvent these challenges by using personal accounts to access AI tools without attracting scrutiny from larger vendors. However, this workaround is unsustainable in the long run, particularly as major AI providers seek to balance their books and demonstrate profitability to shareholders.
Rob Steele, CFO at UK-based accounting software firm iplicit, emphasises the importance of crafting precise prompts to manage costs effectively. “You wouldn’t send someone in your family out to get the weekly shop without any kind of detailed instructions,” he quips, underscoring the need for clarity in AI interactions to avoid unnecessary token expenditure.
Strategies for Pricing AI Services
As companies grapple with these challenges, many are exploring various pricing strategies. Bill Peterson, Senior Director of Product Marketing at Sumo Logic, notes the complexity involved in establishing a pricing model for new services based on agentic AI. “We’re still having some fun conversations about this internally,” he shares, acknowledging the uncertainty that comes with variable pricing structures.
Potential options include raising prices across the board, charging based on outcomes, or offering bundled services. However, these strategies must remain flexible to adapt to the fluctuating costs imposed by LLM providers. “You get into variable pricing, and it’s changing every couple of months. Customers don’t like that,” Peterson adds, pointing out the difficulties of budgeting in such an erratic environment.
Why it Matters
Understanding the intricacies of tokenomics is vital for anyone engaging with AI technology. As businesses increasingly rely on AI solutions, the unpredictability of token costs can lead to budgetary chaos and operational inefficiencies. By grasping these dynamics, companies can better prepare for the future, ensuring that their AI investments yield maximum value without breaking the bank. As the landscape continues to evolve, so too must our approaches to pricing and managing AI services, making this a crucial topic for industry stakeholders.
