There is a mistake investors make in almost every bubble.
They think the question is: “Is the technology real?”
That is usually the wrong question. Railways were real. The internet was real. Smartphones were real. Cloud computing was real. AI is real too.
The better question is uglier and more useful:
“How much of the future has already been priced in?”
That is where the AI story becomes uncomfortable. I do not think artificial intelligence is fake. I use it. Companies use it. Software is changing. Search is changing. Coding is changing. Customer support is changing. Design, data analysis, translation, advertising, finance, education - all of these are being touched by AI in some way.
But a technology can be revolutionary and still become a bad investment at the wrong price. That is the part people forget when markets are euphoric.
A real technology can still create a financial bubble
The dot-com bubble is the obvious comparison, and it is overused because it is useful. The internet did change the world. Actually, it changed the world even more than many people expected in 1999. Online shopping, streaming, cloud software, digital advertising, social media, smartphones, remote work - all of this came from the internet era.
And yet many internet stocks were terrible investments at the peak. Amazon survived and became one of the most important companies on earth. But even Amazon’s stock fell brutally after the dot-com mania. The technology was real. The price was the problem.
That is the uncomfortable possibility with AI. Maybe AI changes everything.
Maybe AI also produces terrible returns for investors who bought the excitement too late. Both things can be true.
Markets do not pay you for being right about the future in a general way. They pay you for buying future cash flows at a reasonable price. If you pay too much, even a correct story can disappoint you.
Why this is not just a stock market story
At first, the AI boom looks like a Nasdaq story. Nvidia, Microsoft, Meta, Alphabet, Amazon, Apple, Tesla, Oracle, data centers, chips, cloud spending, model training, electricity, memory, cooling systems.
But it is bigger than that.
The AI boom has become tied to the US economy, US stock market valuations, US bond yields, the dollar, and global capital flows.
When investors buy the AI story, they often buy US assets. They buy US tech stocks. They buy S&P 500 ETFs. They buy Nasdaq exposure. They buy the dollar indirectly because the biggest AI winners are priced in dollars and listed in the United States.
So if the AI boom continues, it can support the dollar through capital inflows.
If the AI trade cracks, it can work in reverse.
That does not mean the dollar collapses tomorrow. Currency markets are messy. Interest rates, inflation, central banks, energy prices, geopolitics and safe-haven flows all matter. But when one market story becomes large enough, it can start pulling currencies with it.
AI has become one of those stories.
The S&P 500 is not as diversified as people think
Many investors buy the S&P 500 because it feels safe and diversified. Five hundred companies. Different sectors. American economy. Global champions. Simple.
That was always partly true. But it is less true when a small group of mega-cap technology companies dominates the index.
If a handful of companies represent a very large share of the index, then buying the S&P 500 is not just buying “the US market”. It is also buying a concentrated bet on the biggest companies, and today many of those companies are directly or indirectly tied to the AI narrative.
This does not make the S&P 500 bad. It makes it different from what many people think they own.
A tourist may think Europe equals the euro. An investor may think S&P 500 equals diversification. In both cases, the shortcut is useful until it becomes misleading.
When the same few companies drive returns, the index becomes more vulnerable to disappointment in those companies.
And disappointment does not require disaster. Sometimes all it takes is slower growth, lower margins, higher capex, weaker guidance, or the market deciding that "amazing" is no longer enough.
The AI capex race is the part that worries me
The most interesting part of the AI boom is not the chatbot demo.
It is the spending.
Big technology companies are pouring enormous amounts of money into data centers, chips, servers, power, cooling and cloud infrastructure. This is not a small software upgrade. It is an industrial buildout.
That can be bullish. If demand is real and monetization follows, the companies building AI infrastructure may own the rails of the next economy.
But it can also be dangerous. Capital expenditure has to earn a return. Someone eventually has to pay for all those chips, buildings, electricity contracts and cloud clusters. If AI products do not produce enough revenue and profit, the market will start asking harder questions.
At the beginning of a boom, investors reward spending because it signals ambition. Later, they may punish the same spending because it looks like overcapacity.
That is usually how cycles change.
In the early phase, everyone says: "They are investing for the future."
In the late phase, the question becomes: "What if they invested too much?"
Why bond yields matter in the AI story
A bubble does not need low interest rates to exist, but low rates make bubbles easier.
When money is cheap, future profits look more valuable. Investors are willing to pay more for growth that may arrive many years from now. This helps long-duration assets: technology stocks, startups, speculative growth companies.
But when bond yields are higher, the math changes.
If investors can earn a reasonable return from bonds, they become less willing to pay any price for distant future growth. The discount rate rises. Valuations matter again. Promises need to become cash flows.
This is why the AI boom is connected to the bond market.
If US yields stay high, expensive growth stocks face more pressure. If yields fall, the market may become more forgiving again. If yields rise because deficits, inflation or debt worries return, the AI trade may face a tougher environment.
Exchange rates feel this too.
Higher US yields can support the dollar. But if higher yields come with concerns about fiscal stress or overvalued equity markets, the effect can become more complicated. Currency markets do not move from one variable. They move from the argument investors believe that week.
What happens to the euro if the AI trade breaks?
Europe is not the center of the AI equity boom. The big story is American.
That can be both a weakness and an advantage.
Europe has fewer mega-cap AI platform companies. It has ASML, SAP, semiconductor suppliers, industrial automation, energy infrastructure, telecoms, defense, luxury, banks and pharma. But it does not have the same direct AI market leadership as the United States.
In a normal AI melt-up, that can make European equities look boring.
But if the AI trade corrects, boring may suddenly become interesting.
The euro could be affected in several ways. If US tech stocks fall sharply and investors rush into dollars for safety, the dollar may strengthen and the euro may weaken. If the correction is mainly about US overvaluation and capital rotates toward cheaper non-US markets, the euro could benefit.
If the AI boom has been supporting US growth expectations and those expectations fade, the dollar could lose part of its shine.
There is no single automatic answer. That is what makes currency markets difficult.
But the key point is simple: a large AI correction would not stay inside Silicon Valley. It would travel through stocks, bonds, capital flows and exchange rates.
AI winners may exist even if the AI trade disappoints
This is important.
Saying "there may be an AI bubble" is not the same as saying "AI is useless". It is also not the same as saying every AI-related company will fail.
Some companies will make a lot of money. Some already do. The infrastructure layer may produce durable winners. The software layer may create new monopolies. The productivity gains may be real. Some companies that look expensive today may grow into their valuations.
The problem is that investors often treat the whole theme as one trade. AI chipmakers. Cloud providers. Software companies. Power suppliers. Data center REITs. Consulting firms. Startups with no profits. Anything with "AI" in the investor presentation.
They are not the same.
A real boom always contains both future champions and future embarrassments. At the peak, the market struggles to tell them apart. Everything moves together because the story is stronger than the details.
Later, the details return.
Why "this time is different" is partly true and still dangerous
Every bubble has something new. That is why intelligent people get caught in it. If nothing were new, the story would not work.
AI is genuinely different from many older technologies. It can write code, summarize documents, generate images, support research, automate parts of office work, analyze data and interact in natural language. The speed of adoption has been remarkable.
So yes, this time is different. But valuation cycles are not different. Human psychology is not different.
Capital still overbuilds when incentives are strong. Investors still extrapolate recent growth too far into the future. Companies still spend aggressively when they fear being left behind. Analysts still raise price targets after prices have already gone up.
Retail investors still discover concentration risk after the correction, not before.
The technology changes. The behavior is familiar.
What should a normal investor do?
Not panic. Panic is usually a bad strategy. So is pretending valuation does not matter. The reasonable answer is boring: Diversify.
Do not own only one country. Do not own only one theme. Do not assume the last 10 years must repeat. Do not confuse a great company with a great stock at any price.
Do not believe that an index fund is automatically diversified just because it has hundreds of names.
For a European investor, this may mean looking beyond the S&P 500. A global ETF, developed markets plus emerging markets, some bonds, some cash, maybe local-currency assets depending on personal needs. Not because the US is finished. That is too dramatic. The US still has the deepest capital markets, the strongest tech ecosystem and many of the best companies in the world.
The issue is not quality. The issue is price and concentration. A wonderful asset can become risky when everyone owns it for the same reason.
What should currency watchers pay attention to?
For exchange rates, I would watch five things.
First, US bond yields. If yields rise or fall sharply, the dollar will react.
Second, AI earnings. Not just revenue growth, but margins, capex, guidance and whether companies can show that AI spending is turning into profit.
Third, market breadth. If only a few mega-cap stocks hold the market up, the structure is fragile.
Fourth, capital flows. If global investors keep buying US assets, the dollar gets support. If they diversify away, the euro and other currencies may benefit.
Fifth, central banks. The Federal Reserve and the European Central Bank still matter. AI may be the exciting story, but interest rate expectations often decide the currency move.
This is why exchange rates are never just about exchange rates.
They are a summary of many arguments happening at once.
The honest conclusion
I think AI is real. I also think parts of the AI market look like a bubble.
That is not a contradiction. It is usually how major technology booms work. The real thing attracts real capital, then too much capital, then unrealistic expectations, then a correction, then the survivors build the future more quietly.
The hard part is timing.
A bubble can continue long after cautious people start sounding intelligent. Expensive markets can become more expensive. A stock that looks absurd can double before it falls. Selling everything because someone says “bubble” can be just as dangerous as buying everything because someone says “revolution”.
So the useful lesson is not: run away from AI. The useful lesson is: respect the difference between technological truth and investment price.
AI may reshape the economy. It may boost productivity. It may create extraordinary companies. It may also disappoint investors who paid for perfection.
For the dollar, the euro and global exchange rates, the AI boom is now too large to ignore. It influences US equity valuations, bond market expectations, global risk appetite and capital flows.
A chatbot may look like software. But the AI boom has become macro. And once a technology story becomes macro, currency markets start listening.