The development of artificial intelligence is not just a software issue. Behind every new model lies extensive infrastructure—powerful chips, data centers, electricity, networks, and cloud services. That is precisely why AI has also become a story of massive capital expenditures. Big tech companies continue to build out the infrastructure needed to operate and further develop AI. According to current estimates by Bank of America Global Research, Microsoft, Alphabet, Meta Platforms, Oracle, and Amazon could collectively spend nearly USD 800 billion on AI infrastructure in 2026. Gartner estimated that globally, an astronomical USD 2.52 trillion could be poured into this area in 2026, representing a 44 percent increase compared to the previous year. Such numbers naturally raise the question of whether demand for AI services will grow fast enough to deliver an effective return on these investments.
The First Winners Are Already Known
Some infrastructure providers are already benefiting significantly from the boom. One of the most prominent examples is Nvidia. Its graphics processing units have become the core building block of modern AI data centers. However, the company is not the only link in the chain. Manufacturers of memory chips, semiconductor equipment, data storage, and data center operators themselves are also participating in the growth of AI infrastructure. Companies like SK Hynix, Nvidia, and other suppliers demonstrate that the first phase of the AI boom is creating value primarily at the infrastructure level. A big question mark for investors, however, remains whether the companies purchasing this infrastructure will prosper in the long term. [2]
The Most Important Question: Who Will Pay?
Tech companies can invest billions in AI, but investments alone do not guarantee profit. To sustain the entire ecosystem, companies will need to monetize their AI products and services. This means that customers—from large corporations to everyday users—will have to be willing to pay for them. This is where the biggest difference between today's AI boom and the tech revolution at the turn of the millennium may lie. A technology can be successful, yet an investor can still lose money if they pay too high a price for its future potential. [3]
AI Can Create Value, but Also Destroy Margins
Companies invest in artificial intelligence to boost productivity, reduce costs, or generate new revenue streams. However, if prices for AI services fall due to competition, part of the expected profits could disappear. For investors, therefore, it will not only matter how much a company invests in AI, but above all how much it can earn from those investments.
Where Does the Future Potential Lie?
The AI investment story may gradually shift from chip manufacturers themselves to companies that can effectively utilize artificial intelligence. Businesses operating in cloud services, cybersecurity, software, or automation could therefore be interesting. If AI becomes a standard part of doing business—and as we can already see today, it is establishing itself in the market at immense speed—the value may not remain solely with companies supplying computational power. It could also shift toward companies that can translate AI into higher productivity and real revenues. This also explains why, when looking at tech stocks, it pays to look beyond simple revenue growth. Margins, cash flow, capital expenditures, and return on invested capital are what truly matter. [4]
What If the Pace of Investment Slows Down?
Current concerns over a potential slowdown in AI development have already managed to spark investor nervousness, which translated into semiconductor stock prices. In September 2026, companies like Nvidia, Micron, and Marvell came under pressure after representatives of prominent AI firms pointed out the need for a more cautious pace of development. This does not automatically signal the end of the AI boom. SoftBank, which holds investments in OpenAI, also faced a decline in valuation. However, it serves as a reminder that technology stock prices today are heavily dependent on expectations of continued investment growth. If capital expenditures by major tech firms begin to slow, the impact could ripple across the entire supply chain—from chipmakers to data centers. [5]
Bubble or Investment Supercycle?
The facts mentioned above do not mean the AI boom is bound to be a classic bubble. It could be a genuine technological supercycle that reshapes productivity and the operation of entire industries. Still, this does not mean every company connected to AI will automatically succeed. Financial market history shows that even major technological revolutions can create periods where investor expectations outpace fundamentals. The internet was a true revolution, yet the dot-com bubble proved that having the right technology does not guarantee the right stock price. With AI, the key distinction may be separating the companies supplying AI, the companies funding it, and those able to convert it into actual profit. [6]
Investors Should Be Cautious to Choose Wisely
The answer to who will ultimately be the winner and who the loser is not yet clear-cut. Part of the cost is borne by tech firms through massive capital expenditures, part by customers paying for AI services, and part by investors funding the sector's rising valuations. One thing can already be said with certainty: the AI boom will eventually need to generate economic returns. If billion-dollar investments in chips, data centers, and software can be turned into higher productivity and new revenues, today's heavy investments could mark the beginning of a new era of economic growth. However, if expectations grow faster than real profits, the bill for investors who put their capital into the wrong companies could be far higher than it seems today. That is why the most important question about AI might not be whether the technology will change the world, but rather who will own the infrastructure, who will collect the revenues, and who will end up paying for AI growth without an adequate return on investment. [7]
[1,2,3,4,5,6,7] Forward-looking statements are based on assumptions and current expectations, which may be inaccurate, or on the current economic environment, which may change. Such statements are not guarantees of future performance. They involve risks and other uncertainties that are difficult to predict. Results may differ materially from those expressed or implied in any forward-looking statements.
