If one looks attentively, without superfluous emotions, at the technological market of the beginning of 2026, one can sense a strange, I would even say, paradoxical cognitive dissonance. The situation is twofold. On the one hand, we observe incredible, simply cosmic technical progress: models have become smarter, they have become faster and, what is critically important thanks to aggressive price pressure from the side of new ambitious players like DeepSeek, they have become dramatically cheaper. Computational capacities, which a mere five years ago seemed to us distant science fiction, are now available to any student via API literally for pennies. On the other hand, the former unrestrained euphoria in the financial markets is gradually being replaced by cold caution. The reports of technological giants are beginning to provoke more and more uncomfortable questions. Large investors, who have invested trillions of dollars in physical infrastructure (CAPEX), are beginning to ask one and the same unpleasant question: "And where, properly speaking, is that very exponential growth of the world economy which was so confidently promised to us?" As a financial analyst, I am accustomed to trusting dry figures, and not marketing hype. And current macroeconomic indicators unequivocally hint that we have approached a critical fork in the road. The current model of artificial intelligence implementation is hitting a natural, hard economic ceiling. To understand where we should move further, we should introduce a fundamentally new system of coordinates and clearly divide the influence of AI into two fundamentally different categories: Type 1 (Optimization) and Type 2 (Creation). Part 1. Macroeconomic Theory of Value Part 1. Macroeconomic Theory of Value The key to understanding the future lies not in code, but in a deep understanding of the influence of technologies on world GDP, on aggregate demand and aggregate supply. Type 1: The Optimization Trap Type 1: The Optimization Trap The first type is the use of AI in order to make already existing things better, faster, or simply cheaper. This is the automation of routine, the improvement of logistics, the optimization of program code, or the direct replacement of human labor with a program algorithm. In this scenario, AI acts as a "smart tool" or as an efficient replacement for an employee. Technologically this is, undoubtedly, impressive. But economically here lies a treacherous catch, which I call the "Slippers Factory Paradox." Let us imagine a world economy which stably consumes 1 billion pairs of slippers per year. Demand for slippers is stable, humans have only two legs. If one introduces advanced AI and robots into this factory, one can radically reduce costs and increase production efficiency twofold. What happens in the end? The cost price of production falls.
The profit of a specific company grows (in the moment, until competitors have done the same).
Part of the processes becomes automated, ruthlessly displacing human labor. The cost price of production falls. The profit of a specific company grows (in the moment, until competitors have done the same). Part of the processes becomes automated, ruthlessly displacing human labor. But the most important thing, about which everyone forgets: The world does not begin to buy 2 billion pairs of slippers just because they became easier to produce. Aggregate final demand does not grow. The market is limited by the physical need of a human being. From the point of view of world GDP, the total pie has not increased by an inch. Only a redistribution of market shares and cannibalization has occurred: more efficient companies (Type 1) simply displaced and killed less efficient ones. This is a classic zero-sum game. It is in this phase that we, in essence, have been for the last two years. We optimized search, we optimized content, we optimized code. We squeezed efficiency out of the old economy to the last drop. But efficiency has a limit — one cannot reduce costs below zero. Type 2: Creation of New Value Type 2: Creation of New Value The second type is a completely different story. This is the use of AI for the creation of goods and services which physically did not exist in nature before the appearance of this technology. These are products which do not replace old consumption, but create fundamentally new consumption. Let us recall the appearance of smartphones. After all, this was not simply an "improvement of the push-button telephone." No. This created a gigantic industry of mobile applications, taxi services, video streaming, and social networks in a pocket. People began to spend money on services which earlier were simply not on humanity's "menu." Type 2 expands the nomenclature of goods. It grows real GDP both by demand and by supply. This is the only form of growth which is capable of justifying the current colossal, trillion-dollar investments in infrastructure. Part 2. Current Market Landscape Part 2. Current Market Landscape If one looks at the leaders of the industry through the prism of this theory, the picture becomes much clearer and more dramatic. Google (Alphabet): The company stands before a classic, textbook "innovator's dilemma." Their main business — Search — is pure Type 1. The introduction of generative AI improves the quality of answers, yes, but it increases their cost price. This is the protection of current positions, the defense of a fortress, and not the creation of a new market.


Microsoft: Their global bet on Copilot is a bet on productivity (Type 1). This is a huge market, there is no dispute, but it is finite. The sale of "work acceleration" has a natural saturation limit. One cannot work 25 hours a day.


Meta Platforms: The strategy of open code (Open Source) turns advanced AI models into a publicly available resource (commodity). This lowers entry barriers for all startups, but simultaneously destroys super-profits in the sector of base models. Mark Zuckerberg is playing the long game, burning the margin of competitors.


NVIDIA: The main beneficiary of the current moment. King of the hill. But here there is a hidden risk: demand for their chips is derivative. Companies purchase GPUs to engage in optimization (Type 1). If the economy of optimization hits the ceiling of profitability, demand for infrastructure may correct harshly. Google (Alphabet): The company stands before a classic, textbook "innovator's dilemma." Their main business — Search — is pure Type 1. The introduction of generative AI improves the quality of answers, yes, but it increases their cost price. This is the protection of current positions, the defense of a fortress, and not the creation of a new market. Google (Alphabet): The company stands before a classic, textbook "innovator's dilemma." Their main business — Search — is pure Type 1. The introduction of generative AI improves the quality of answers, yes, but it increases their cost price. This is the protection of current positions, the defense of a fortress, and not the creation of a new market. Google (Alphabet): Microsoft: Their global bet on Copilot is a bet on productivity (Type 1). This is a huge market, there is no dispute, but it is finite. The sale of "work acceleration" has a natural saturation limit. One cannot work 25 hours a day. Microsoft: Their global bet on Copilot is a bet on productivity (Type 1). This is a huge market, there is no dispute, but it is finite. The sale of "work acceleration" has a natural saturation limit. One cannot work 25 hours a day. Microsoft: Meta Platforms: The strategy of open code (Open Source) turns advanced AI models into a publicly available resource (commodity). This lowers entry barriers for all startups, but simultaneously destroys super-profits in the sector of base models. Mark Zuckerberg is playing the long game, burning the margin of competitors. Meta Platforms: The strategy of open code (Open Source) turns advanced AI models into a publicly available resource (commodity). This lowers entry barriers for all startups, but simultaneously destroys super-profits in the sector of base models. Mark Zuckerberg is playing the long game, burning the margin of competitors. Meta Platforms: NVIDIA: The main beneficiary of the current moment. King of the hill. But here there is a hidden risk: demand for their chips is derivative. Companies purchase GPUs to engage in optimization (Type 1). If the economy of optimization hits the ceiling of profitability, demand for infrastructure may correct harshly. NVIDIA: The main beneficiary of the current moment. King of the hill. But here there is a hidden risk: demand for their chips is derivative. Companies purchase GPUs to engage in optimization (Type 1). If the economy of optimization hits the ceiling of profitability, demand for infrastructure may correct harshly. NVIDIA: Part 3. The Factor of Intelligence Deflation Part 3. The Factor of Intelligence Deflation Transition to Type 2 becomes not simply an ambition or a dream, but a harsh necessity due to the factor of competition. In the years 2025–2026, we saw that the cost of creating and using intellectual models is falling rapidly. The appearance of powerful and cheap models (for example, from Chinese developers like DeepSeek) means inevitable deflation of intelligence. The cost of a "smart token" strives toward mathematical zero. To build a business which is simply a "wrapper" over someone else's model and performs the function of banal optimization becomes economically senseless. Margin in the Type 1 sector evaporates before one's eyes. Competition turns into a bloody "race to the bottom." Part 4. The Mirage of Autonomous Agents Part 4. The Mirage of Autonomous Agents Many hopes, possibly even too many, are now placed on Autonomous Agents. But it is important to understand a fundamental thing: if an Agent simply replaces an employee (for example, an accountant or a tourism manager), this is still Type 1 economy. This changes the cost structure of corporations, this is profitable for business, but this does not necessarily create new value for the world as a whole. Roughly speaking, if an algorithm performs a human's work, the global quantity of produced goods does not change. This is a redistribution of income, and not organic growth. The true potential of Agents will be revealed only when — and not a minute sooner — they begin to do what was inaccessible or too expensive for people. Part 5. Golden Time for Seeking Ideas Part 5. Golden Time for Seeking Ideas We are located at a unique historical point — in the Investment Phase. Ten years ago technological giants sat on mountains of cash and literally did not know where to invest it. Today the situation is mirror-opposite: capital is actively, even aggressively spent on the construction of data centers and energy infrastructure. Capital has ceased to be a deficit. Ideas have become the deficit. Infrastructure is being built at Breakneck speed. The "railroad" is being laid. But the "trains" (Type 2 products) are yet to be invented. And it is precisely here, in this gap, that a window of opportunity opens for developers and entrepreneurs. What is a Type 2 product? It is that which expands the boundaries of consumption. Let us look at history. The growth of the economy over the last 500 years is the growth of the nomenclature of goods. We consume thousands of things about which our ancestors did not even know and did not dream. The task of the current generation is to add new items to this consumer basket. Example 1: A New Intellectual Good Example 1: A New Intellectual Good Let us consider the concept of an "Intellectual Partner." Quite recently such a service physically did not exist. A person could either hire a live assistant (expensive, difficult), or seek information himself (long, tedious). The appearance of the possibility to conduct a dialogue with a competent AI 24/7 created a completely new market. People are ready to pay for this not as for a replacement of something old, but as for a new entity improving the quality of life. This is pure Type 2. Example 2: Robotics and the World of Atoms Example 2: Robotics and the World of Atoms While AI remains locked in the digital world, its influence is limited by the monitor screen. An exit into the physical world of atoms is the creation of a new labor force. Speech is not necessarily about complex humanoid robots. Speech is about machines capable of performing tasks which a human does not want to perform or physically cannot (work in dangerous radioactive environments, microsurgery, maintenance of infrastructure in open space). This creates economic activity where earlier there was a vacuum. Example 3: Technologies Can Be Simple Example 3: Technologies Can Be Simple A Type 2 product is not obliged to be based on the most complex code. The secret is not in the complexity of technology, but in the finding of a new sales market. Apple created the tablet market by simply rethinking the form factor. Sometimes a great innovation consists simply in finding an empty niche in human needs and filling it. Conclusion Conclusion The artificial intelligence market is maturing before our eyes. The period of easy, crazy growth on hype is ending, yielding place to harsh macroeconomic reality. Type 1 Economy (Optimization) will be tough, bloody, and low-margin. There, giants and free Open Source will compete to the death. The real opportunity, the "blue ocean," today lies in the plane of Type 2. Humanity possesses excess computational capacities and capital ready to finance breakthroughs. The task consists not in optimizing the old world a little bit more, by half a percent. The task is to invent what to fill the new one with. Ideas are needed. Ideas which will create markets not existing today. And, judging by appearances, it is precisely on the ability to find these ideas that the entire economic success in the coming decade will depend.

The is an opinion piece based on the author’s POV and does not necessarily reflect the views of HackerNoon.

Microsoft

Why AI Optimization Is a Dead End for the Economy

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

Where's the AI Revenue? Plus, Will the Trillion-Dollar AI Gamble Pay Off?

Where's the AI Revenue? Plus, Will the Trillion-Dollar AI Gamble Pay Off?

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps