The world competition in AI has taken an unexpected turn. Whereas companies from all around the globe have been pushing AI adoption with vigor during the past couple of years, now firms like Microsoft and Uber find themselves facing a disturbing discovery—AI technology might prove much more costly than expected, and in certain situations, it is even less efficient than human labor.
This situation becomes apparent as corporations spend huge sums of money on implementing various solutions based on AI. In particular, businesses invest billions of dollars in generative AI, automations, and coding tools based on AI.
Microsoft Cancels Licenses for Its AI Tools
As noted in recent news reports, Microsoft is now canceling direct licenses for its own AI-based coding assistant called Claude Code. The company advises users to switch to the tools created internally, specifically GitHub Copilot CLI.
What is particularly interesting about this news story is that just a few months earlier, Microsoft was actively promoting the use of AI tools for coding among thousands of developers, project managers, and designers. As soon as the implementation took off, however, it caused a spike in expenses.
It is not that the technology does not work. What has happened is that the cost structure needed for running complex AI models has become very expensive for many businesses, a cost that they underestimated.
While reducing its use internally, it seems that Microsoft’s partnership with Anthropic continues unabated, including huge cloud infrastructure deals worth billions of dollars through Azure.
Uber Says AI Spending is Difficult to Justify
Ride-sharing company Uber has also reported issues regarding AI spending. According to reports, Uber’s management has reported that the company blew through its budget allocated for using AI tools until 2026 in just four months.
Earlier, Uber had tried to encourage the use of AI technology amongst its workforce by setting up leaderboards based on the use of these tools. However, senior officials at Uber later found out that despite increasing AI spending, their productivity remained the same.
According to a statement by the Uber COO Andrew Macdonald, “The productivity gains from the spend weren’t there.”
The reason why this point is important is because of the big promise that AI had made – it would slash labor costs while increasing production.
The Underside Issue: AI Token Economics
A significant reason behind the surge in AI costs can be explained through the idea of token economics.
Many state-of-the-art AI services bill businesses for using “tokens”—fragments of text used as inputs in AI model operations. More use results in greater consumption of tokens and an increase in the associated costs.
Interestingly, businesses pushed employees to use AI as much as possible, thinking that it will enhance productivity. Instead, they incurred exorbitant costs due to their computational bills.
According to the reports, some prominent tech firms such as Amazon and Meta internally promoted “tokenmaxxing,” a phenomenon whereby employees were incentivized to use as many tokens as possible.
Furthermore, agentic AI systems, which perform tasks that require multi-step reasoning, require a lot more computing power than regular AI chatbots do.
Productivity Benefits of AI May Be Exaggerated
Several recent pieces of research and market reports point out that organizations may be exaggerating the impact of AI on productivity.
Recent research shows that users tend to perceive that AI saves more time than it really does. In most cases, work aided by AI did not demonstrate any considerable speedup compared to regular work.
This effect has been referred to as the “speedup illusion,” in which employees feel more productive in the use of AI technologies despite minimal performance improvements.
This poses an investment return-on-investment problem to those corporations that have to pay millions of dollars yearly in the development of AI infrastructure.
Why Is AI Critical for Big Tech?
Nonetheless, the use of AI is expected to stay among companies.
As per industry experts, the importance of AI in areas such as software development, customer support, data analysis, and automation is not going anywhere.
Businesses would likely adopt a mixed strategy in which AI helps people in their work instead of fully replacing employees.
Currently, AI functions as a productivity tool rather than a replacement for human labor due to some reasons.
For instance, in sectors such as artistry, judgment, decision-making, and emotional intelligence, AI cannot compete with people.
AI agents are expected to raise business expenditures even further.
According to the investment bank Goldman Sachs, AI agents can increase token utilization by as much as 24 times by 2030.
While it is expected that the price per token will eventually fall, the increased volume of AI use in organizations may lead to an increase in enterprise spend on the technology.
The research firm Gartner also cautioned that even as AI technologies become less expensive, enterprise-level AI costs will still not drop since advanced reasoning platforms consume far greater compute power.
What Does It Mean for the Future of Work?
Recent developments from Microsoft and Uber indicate that the automation of the workplace may take longer than originally anticipated.
It appears that rather than completely replacing humans, organizations will increasingly seek to leverage the advantages of human-machine collaboration, recognizing that blind adoption of AI can prove costly without tracking results.
For the employee, this implies that previous fears about mass automation may have been somewhat unfounded, as human capabilities such as intuition and judgment are highly valued.
At the same time, workers who embrace collaborative work with intelligent machines will find themselves at an increasing advantage going forward.
Conclusion
The age of AI is not yet over, but it can certainly be seen that the narrative is changing. The examples of Microsoft and Uber show that implementation of AI cannot just rely on its capabilities; there needs to be some level of sustainability.
Even though AI continues to be one of the most revolutionary developments of the decade, companies are beginning to realize that AI entails high costs, uncertain productivity levels, and complex processes.
The future race of AI may shift from replacing humans to striking a perfect balance between AI and human intelligence.
