Tokenization: Companies now know the cost of AI—but can’t measure the value

Tokenomics shock recently swept through C-suites as vendors shifted from seat-based licenses to token-based artificial intelligence pricing. Organizations finally realized the exact cost of their AI usage. But would their reaction to those seemingly excessive costs be the same if the invoice were for a $100,000-a-year employee? Probably not.  

While employee compensation is a familiar business practice, most organizations still view AI as a traditional software expense. However, AI now plays the role of a digital contributor that can create content, analyze data, make decisions and execute tasks.  

Leaders who—for decades—have measured AI as digital labor with the same metrics used for technology are now calculating value inaccurately. Instead of reducing AI use, tokenization forces businesses to create a new operating model to measure, govern and deploy AI systems. 

AI is not software—it's labor 

With organizations’ visibility into the high costs of AI, many overcorrected by shifting from broad AI use to an intense focus on costs. Additional AI spending is not inherently a problem. But many businesses measure token consumption while prioritizing productivity without defining AI’s role in the workforce.  

Organizations do not evaluate human employees based solely on how quickly they work. They typically prioritize quality, outcomes and business impact. The same mindset must apply to AI. Executives should be fine with spending $1 on a token if it creates $2 of value. When leaders treat AI as digital labor, they keep the focus on value instead of costs, just as they do with human labor.  

As AI use initially soared and the technology quickly became critical for growth, many organizations devoted significant resources to chatbots and agents without measuring the return on investment for each use case. As a result, the first step in the transition to a value-based mindset required for tokenomics starts with a review of current AI use cases and an assessment of changes needed to support a new operating model. 

Recognition of AI as digital labor is only the beginning. The bigger challenge is the development of an operating model that allows organizations to measure, govern and scale it effectively.  

When leaders treat AI as digital labor, they keep the focus on value instead of costs, just as they do with human labor. 

The cost question is the wrong question   

Once leaders agree that a value-based mindset toward digital labor is the right move, creating processes to measure AI capabilities often becomes the next roadblock. The first step is to understand the company’s application stack, especially which applications are token-based and which are subscription-based. Then determine how to make the most of applications that do not use tokens.  

Next, organizations need to set aside the cost calculation as the primary measure and create a new measurement process. Because traditional software metrics are not accurate for the AI workforce, it’s tempting to create a new set of metrics. But organizations can use other existing metrics to most efficiently get a picture of the impact of AI on the entire business.

The RSM Middle Market AI Survey 2026: U.S. and Canada found that 99% of organizations measure ROI for AI. Organizations reported measuring process efficiency (40%), productivity and time savings (38%) and decision quality or speed (36%). However, these metrics do not give a true picture of the business value that digital labor brings to the organizations. 

These key metrics are key mechanisms to measure the value of digital labor to the organization:

While organizations should track tokens for budget purposes, consumption isn’t a true business metric. When organizations focus on how AI usage improves existing metrics, organizations can justify investment and AI adoption at a greater scale.

While organizations should track tokens for budget purposes, consumption isn’t a true business metric.

Governing a workforce you can’t see

Governance is not just about controlling software and cutting costs; it also involves oversight of a workforce that never sleeps, scales instantly and makes autonomous decisions. Governance serves as an accelerator for business opportunities instead of the brake that many often assume. If organizations start with a focus on costs, they may cut the wrong ones—and limit business growth. However, once the right governance is in place, reduced costs typically quickly follow.

However, one in three organizations are not focusing on governance for AI projects, reported the RSM Middle Market AI Survey 2026: U.S. and Canada. Within that 33%, the survey found that 16% of respondents only address governance after issues arise and 3% do not use governance at all.

Governance means a lot of different things, so each organization should define governance for its own purposes and culture. With regard to tokenization, businesses should focus governance on individual AI use, such as user behavior, model choice and usage controls. Once the right governance tools and processes are in place for token consumption, organizations must still manage and monitor the costs. Even with governance, the costs can quickly go high if not monitored.

Instead of building an AI governance framework from scratch, organizations can easily adopt existing guidelines from respected organizations such as NIST and ISO, and also leverage comprehensive frameworks such as the RSM AI Governance Framework, which combines best-practice elements of several leading resources. These existing frameworks commonly focus on core pillars such as strategy alignment, risk management, ethics and accountability, regulatory compliance, and ongoing monitoring. As a result, the frameworks create a flexible governance approach that evolves to changing technology and regulations. 

Preparing the human workforce for digital labor   

Companies that master the technology and measurement aspect of AI as digital labor but do not educate the organization as a whole leave many opportunities on the table. AI is unique because it cuts across organizational silos, which creates challenges that previous technology transformations often did not. Key departments that are involved include sales, marketing, operations, human resources, finance and legal. 

The shift to AI as digital labor must include change management, especially the idea that AI should be viewed as a new form of labor rather than simply another technology tool. Organizations must undergo “The Great Unthink” where they step back and evaluate current business processes instead of using AI for everything. By rethinking their current business models to see how the business runs differently with AI, and redesigning those processes that passed the test, organizations can then change the model to match the design.  

The foundation of change management for improving AI value starts with training, especially upskilling. Many employees use excessive tokens through incorrect AI use, which reduces tokens’ value. Other employees make errors that lower value through low quality of work. Additionally, organizations often see lower value when employees choose a tool that is too powerful (and too expensive) for the task at hand.  

By first determining where employees are in their personal AI journey, businesses can create training that meets their current needs. When training is too broad, employees may not see how it applies to their role and may not pay attention to the rest of the information. Because employees often say that AI feels like a foreign language, each team member should first understand AI terminology.  

The organizations that move first will be positioned to turn digital labor into a competitive advantage rather than simply another technology investment.