Return on Intelligence: The New Definition of ROI in the Age of AI

CULTURE & CODE

When Generative AI entered the mainstream in late 2022, organizations around the world saw something extraordinary happen.

For the first time, exponential computing power appeared to become instantly available, infinitely scalable, and remarkably affordable. Work that traditionally required hours — or even days — could suddenly be completed in minutes. Thousands of documents could be analyzed almost instantly. Software could be developed faster. Customer questions could be answered continuously.

The promise was powerful:

Higher productivity. Greater efficiency. Faster innovation. Lower cost.

It was easy to understand why organizations accelerated adoption. AI appeared to introduce an entirely new kind of workforce — one capable of operating continuously, processing enormous amounts of information, and expanding human capacity in ways previously impossible.

In many ways, organizations believed they had discovered the ultimate employee – one who never sleeps.

Managing the Employee Who Never Sleeps: Agentic AI and the Next Frontier of Leadership

But as AI moved from exciting experiments into enterprise-wide deployment, organizations are now discovering a more complicated reality.

AI may be incredibly powerful, but it now increasingly does not come FREE.

The same technology originally embraced for cost savings and productivity improvement is now creating deeper conversations inside boardrooms about sustainability, governance, and value creation.

More and more, AI at scale also requires financial scale to maintain. Every intelligent response depends on massive computing infrastructure: specialized processors, data centers, energy consumption, cloud capacity, security, integration, and continuous improvement.

Now, the conversation is shifting from “What can AI do?” to “What value does AI create compared to what it costs?”

This becomes increasingly important as organizations deploy AI across thousands of employees and millions of daily interactions. A growing concern, however, is the behavior described as “Token Maxing,” where AI consumption itself becomes interpreted as a measure of productivity.

The intention is understandable. Leaders want employees to embrace AI. Organizations want adoption. Employees want to demonstrate that they are adapting to the future.

But adoption without value creation creates a familiar organizational problem – “Activity” starts looking like “Achievement.”

And companies are beginning to recognize an important lesson:

The employee who never sleeps still sends an invoice.

Enter the Skyrocketing Costs of Token Maxing

“Token Maxing” is a growing workplace trend where employees excessively use AI tools to artificially inflate their performance metrics and appear more efficient.

As AI computational tasks utilize “tokens” which are correspondingly “billed” by the AI service provider to the user organizations (or individual users) – the utilization costs for these have increased. While this consumption-based model initially seemed highly cost-effective, the overall utilization costs for businesses have indeed skyrocketed over time.

The phenomenon began when companies started measuring AI performance and usage by tracking the number of tokens their employees consumed. Consuming a massive amount of tokens quickly became, sort of, a “status symbol” among workers, used as proof that they were going “above and beyond” to meet their key performance indicators (KPIs). As a result, employees — particularly software engineers — started using AI for even insignificant tasks simply to drive up their token count.

This behavior was further encouraged by tech executives who expected engineers to consume hundreds of thousands of dollars’ worth of tokens. Ultimately, this artificial demand severely impacted corporate AI budgets, as generating text, images, videos, and code has now become increasingly expensive and contributed to a significant worldwide increase in overall token costs.

Welcome to the Economics of Intelligence

For decades, organizations understood the economics of traditional software. Build or purchase a system. Deploy it across the enterprise. Scale adoption.

The technology investment was significant, but once established, additional usage often came with relatively manageable incremental cost.

AI introduced a different equation. It behaves less like traditional software and more like a consumption-based resource. Every prompt, analysis, generated image, report, recommendation, and autonomous agent workflow requires computing power. AI does not simply sit inside a server waiting to be opened. It keeps working – and work consumes more resources.

This is why tokens are becoming an important concept beyond technology teams. They represent more than a technical measurement. Increasingly, they represent the economics of digital work.

As major AI companies like Anthropic and OpenAI prepare for Initial Public Offerings (IPOs) in the coming months, future token prices are also expected to multiply exponentially.

Currently, leading AI companies have been operating at a loss, largely subsidizing the real cost of tokens with generous funding to capture market share. However, once these companies go public, they will have to answer to shareholders and face immense investor pressure to generate actual profits and show returns. To achieve this pivot to profitability, they will inevitably increase the prices of using their tokens.

Just as organizations learned to manage employee productivity, technology investments, and cloud consumption, they must now learn to manage intelligence consumption.

Because the goal is not unlimited AI usage. The goal is valuable AI usage.

When AI Adoption Becomes AI Activity

One of the most fascinating challenges of AI transformation may not actually come from technology.

It may come from human behavior.

Organizations have always shaped behavior based on what they choose to measure. When companies rewarded hours worked – people stayed longer. When responsiveness became the standard – email activity exploded. When meeting participation became a sign of collaboration – calendars became overloaded.

AI introduces a new version of the same challenge.

If organizations measure AI transformation through usage alone, they may unintentionally create more consumption instead of more contribution. More prompts do not automatically mean better thinking. More generated content does not automatically mean better communication. More AI agents do not automatically mean better organizations.

A company can proudly report thousands of AI users, millions of tokens consumed, and hundreds of automated workflows running.

But leadership must still ask:

• Did our customers experience something better?
• Did our employees become more capable?
• Did we make faster and smarter decisions?
• Did we create innovation that matters?
• Did business performance actually improve?

Without those answers, organizations risk creating a new form of “productivity theater.”

Only this time, the theater runs on tokens.

The End of the AI “Free Trial” Era

The early years of generative AI benefited from extraordinary investment, rapid innovation, and aggressive market expansion.

Like many breakthrough technologies before it, the first stage focused on adoption:

• Get people experimenting.
• Get organizations exploring.
• Get the world imagining what is possible.
• But every technology eventually matures.

Cloud computing followed a similar journey. Organizations celebrated the ability to access unlimited computing power instantly — until they discovered that unlimited consumption without governance could create very real financial consequences.

The answer was not to abandon cloud. The answer was to become smarter at managing it.

AI is entering the same evolution.

The next stage of AI maturity will not simply be about deploying more tools, more agents, or more automation.

It will be about discipline.

• Where should AI be applied?
• Where does it create meaningful advantage?
• Where does human capability still create greater value?

The next generation of AI leadership will move from measuring adoption to measuring contribution.

From usage – To impact.

Return on Intelligence: The New Leadership Metric

For decades, executives have focused on ROI:

Return on Investment.

Every major business decision eventually comes back to one fundamental question:

“What value did we create compared to what we invested?”

We measure the productivity of capital. We measure operational efficiency. We measure workforce performance.

But AI introduces something organizations have never managed at this scale before – Intelligence itself has become an enterprise resource.

Historically, intelligence was almost exclusively tied to people. Organizations competed by hiring great talent, developing strong leaders, building effective teams, and creating cultures where ideas could flourish.

That remains essential.

However, AI introduced a new dimension. Intelligence can now be generated, expanded, and deployed digitally across thousands of workflows at unprecedented speed. When intelligence becomes abundant, competitive advantage no longer comes simply from having access to it. It comes from knowing how to apply it.

This requires a new leadership meaning on ROI:

Return on Intelligence.

The question is no longer:

“How much AI do we have?”

The better question is:

“How much value does our combined intelligence create?”

The winners of the AI era will not necessarily have the largest models, most AI agents, or highest usage statistics.

The winners will learn how to orchestrate three forms of intelligence:

• Human Intelligence — creativity, empathy, ethics, experience, and judgment.
• Artificial Intelligence — speed, scale, automation, analysis, and pattern recognition.
• Collective Intelligence — the organizational capability created when humans and machines learn, adapt, and create together (SUPERTEAMS).

The future leader is no longer only a manager of people, processes, and technology. It will be those who become architects of organizational intelligence.

The Future Is Not Humans vs. Tokens

Companies like Uber are now beginning to reconsider and slow down their aggressive AI adoption, primarily because the cost of operating AI has skyrocketed – to the point where it can now exceed the cost of human labor.

Several specific factors are driving this reevaluation

• Unsustainable Expenses and Budget Exhaustion: Major tech companies are finding it nearly impossible for executives to justify the massive expenses associated with AI, with some employers blowing through their entire annual AI budgets in just a matter of weeks.
• AI Outpricing Human Workers: While AI remains highly cost-efficient for complex tasks like coding, the financial benefits diminish in other areas. For example, the cost difference between AI and humans is minimal for data entry tasks, and for operations like call centers, hiring human workers is again slowly proving to be cheaper than running AI agents.
• Inflated Token Costs: Token maxing and a massive surge in global demand have artificially inflated the consumption and price of AI tokens. This has exacerbated the financial burden on corporate finance departments, causing companies like Uber, Microsoft, Amazon, and even Nvidia to scramble to slow down their usage.

Ultimately, these spiraling costs have brought companies to a tipping point where they must question the true return on their investments, forcing them to decide if it makes sense to pay more for AI tools than they previously paid the human employees they replaced

As these AI economics become more visible, it is tempting to reduce the future of work into a simple equation:

Humans cost salaries.

AI costs tokens.

Choose whichever is cheaper.

But that misses the bigger transformation point.

The future of work is not about replacing one cost structure with another. It is about redesigning how intelligence creates value.

The better question is not:

“Who costs less?”

The better question is:

“Which intelligence creates the greatest impact?”

AI will create tremendous advantages where speed, scale, consistency, and computational power matter most.

Humans will continue creating unique value where judgment, creativity, empathy, ethics, relationships, and leadership matter most.

AI can generate possibilities. Humans create meaning.

AI can identify patterns. Humans understand context.

AI can optimize decisions. Humans define what matters.

The organizations that win will not simply replace people with technology.

They will redesign work so people and technology make each other better.

From Artificial Intelligence to Intelligent Leadership

The first era of AI was about ACESS.

The second era was about ADOPTION.

The next era will be about WISDOM.

Because the Future of Work advantage will not belong to organizations that consume the most intelligence.

It will belong to organizations that create the most value from the Return on Intelligence.