It has been one of the most discussed technologies of the decade. Ever since the introduction of next-generation generative AI technologies, firms all over the globe have been scrambling to allocate billions of dollars for building out artificial intelligence capabilities. The promises made were plentiful – AI was going to revolutionise productivity, decrease operational expenses, replace menial jobs with automation, and even generate new opportunities for doing business.

Yet, by 2026, the picture looked quite different. Even though the technology has continued to evolve at an incredible pace, many firms are struggling to realise the full economic potential of AI advancements. More prudent investors are becoming sceptical, more firms are facing mounting pressure from stakeholders to justify their expenditures, and general public opinion regarding AI has taken a less positive turn.

Again, it is not about the lack of potential. It has just proven more difficult to realise economic benefits from the technology.

The ROI Problem

The major issue related to the implementation of artificial intelligence lies in the lack of evidence concerning ROI. Companies continue to invest millions in implementing AI, but according to many executives' reports, the financial outcomes are far from spectacular.

For example, research shows that many organisations are incapable of identifying tangible results from the use of AI in their operations. Moreover, Gartner estimated that at least 30% of generative AI implementations will fail to prove their concept due to such reasons as poor data quality, unidentifiable value for the company, rising expenses, and lack of risk management.

This problem is characteristic of many AI projects. While demonstrating great success at the stage of trials or during a small-scale test run, many systems turn out to be not that efficient when deployed throughout the whole organisation. Enterprises realise that it is much harder to implement AI within their work process than they thought.

Thus, many companies spend money on the implementation without obtaining the desired productivity improvement.

The Economics Are Becoming Difficult to Ignore

Another factor behind the loss of AI's value proposition for businesses is the very high price tag required to run advanced AI technologies. Large-scale AI models require a considerable amount of infrastructure. Tech corporations have spent billions in constructing data centers, specialized hardware, cooling equipment, and electricity. Although there is still high demand for AI-based services, the income of various AI applications still does not justify the investments made.

There has been increased scepticism regarding the sustainability of this business model. At first, many AI organisations had a pay-per-month structure, requiring users to pay a fixed price per month regardless of how much AI they used. However, some heavy users use more computing power than the monthly fees can cover. Therefore, some organisations have begun using pay-per-use billing, based on the amount of computing AI users employ.

The shift demonstrates one simple fact. AI is very costly. While the technology might be amazing, it has to start turning a profit for organisations.

Infrastructure and Physical Limitations

While there are numerous talks about AI and software developments, there is less discussion about the physical foundations necessary for these achievements.

The development of AI relies on massive amounts of electricity. Computing facilities need constant power supplies, cooling, and connectivity with the electrical grid. In numerous regions around the world, the infrastructure needed to provide electricity to computing facilities is failing to adapt to the fast-developing field. Thus, a new thermodynamic bottleneck arises for the development of artificial intelligence. The issue here is not the algorithm that should be created; it is the amount of electricity to be provided to a particular facility where computations take place. It makes the process of constructing AI facilities slower due to physical limitations of electricity generation, its distribution channels, and construction itself.

Even the organisations that are ready to spend billions cannot do so instantaneously.

Rising Interest Rates and Investor Pressure

Financially, the situation is becoming more challenging for artificial intelligence enterprises. It is common knowledge that when interest rates are low, it is easier for companies to get financing for big projects that promise huge returns in the distant future. It was a time of great benefits for artificial intelligence, since money was easy to earn. However, changes in the financial market have altered investor sentiment, since high interest rates significantly decrease the present value of expected income. Consequently, technology companies, which rely mainly on future earnings, come under closer scrutiny.

In fact, according to analysts, many of the problems that occurred recently with artificial intelligence were provoked by this trend rather than by negative attitudes towards the technology. The point is that investors do not trust optimistic forecasts anymore; now they look for tangible results.

Therefore, some companies had to switch to a different approach to work.

From Possibility to Profitability

However, shifts are happening in the discourse about AI. At the early stages of the artificial intelligence boom, all conversations revolved mainly around its potential. Organisations presented stunning examples of its possibilities, its futuristic uses, and the revolutionary transformations it can facilitate. The investors seemed ready to reward these possibilities.

These days, though, things have changed considerably.

Organisations are expected to be able to translate the AI possibilities into tangible revenues.

What people want to know is whether AI can make more money for businesses, cut expenses, improve the customer experience, or create a competitive advantage. There are many organizations which discover that simply adopting AI does not work out. Successful AI adoption implies, in most cases, re-designing workflows, re-structuring the company, and changing business processes.

According to industry experts, the main obstacle standing behind successful AI adoption is, quite often, not the technology itself, but the inability of companies to adapt to it.

Growing Public Scepticism

On the other hand, public perceptions of AI technologies are increasingly becoming sceptical.

According to surveys, a considerable proportion of Americans is apprehensive about the growing role of AI technologies in everyday life rather than enthusiastic about their prospects.

Concerns include fears of automation of jobs, propagation of fake news, invasion of privacy,

and erosion of essential human skills. As found by the research conducted by Pew, a majority of Americans perceive the threats from AI technologies as serious, whereas significantly fewer individuals believe their potential is greater than the risks. Some are worried about the possible harm that AI could do to creativity, critical thinking, and interactions between people.

This growing scepticism is important in light of the impact that public opinion could have on regulations. The need to respond to the voters' concerns could make governments enact legislation aimed at regulating the development and implementation of AI technologies. Even though such legislation could increase the efficiency of these processes, it could also make them more costly.

Looking Beyond the Hype

The truth of the matter is that artificial intelligence still stands as one of the most powerful technologies available in the current age. However, what matters now is not whether AI can accomplish remarkable feats but whether firms can transform those feats into profitable ventures.

As the year 2026 progresses, the emphasis in terms of AI development is increasingly shifting away from the innovation part and towards implementation. Organisations have found that applying AI technology proves far trickier than proving its capability; factors such as high costs, unclear returns on investment, inadequate infrastructure, increased pressure from investors, and public doubt make implementation an increasingly necessary strategy.

Firms that end up succeeding in the coming years might not have been the ones investing in the best artificial intelligence technology, but rather the ones implementing it properly. Despite further developments in terms of technology, businesses are learning that innovation alone is not enough; the bottom line lies in making sure that huge investments bring sufficient financial returns.

References

1. Gartner. 30% of Generative AI Projects Will Be Abandoned After Proof of Concept by the end of 2025. (2024).
2. Pew Research Centre. How Americans View AI and Its Impact on People and Society. (2025).
3. Pew Research Centre. How the US Public and AI Experts View Artificial Intelligence. (2025).
4. Gallup and Bentley University. Americans Express Real Concerns About Artificial Intelligence. (2024).
5. TechRadar Pro. Why Enterprise AI Stalls and What Executives Must Do Differently. (2026).
6. Mercer Executive Survey on AI Adoption and Workforce Impact. (2026).
7. Axios. Americans Are Using and Worrying About AI More Than Ever, Survey Finds. (2025).
8. Grace, K. et al. Thousands of AI Authors on the Future of AI. arXiv. (2024).

.    .    .