Norway’s sovereign wealth fund is making billions from the AI boom. Its own CEO is now warning that the same boom could become a serious source of risk.
For more than two years, artificial intelligence has been one of Wall Street’s most powerful investment stories.
Nvidia, Microsoft, Amazon, Alphabet and other technology giants have poured enormous amounts of money into chips, data centers and AI infrastructure. Investors have followed, pushing technology valuations to extraordinary levels.
But now an uncomfortable question is becoming harder to ignore:
What happens if the money being invested in AI grows faster than the profits AI can ultimately generate?
That question has moved from market speculation into the offices of some of the world’s biggest investors.
Nicolai Tangen, CEO of Norway’s Government Pension Fund Global, has warned that elevated AI valuations represent a significant future risk. His warning is particularly striking because the fund itself has benefited enormously from the technology rally.
And on August 18, semiconductor stocks provided a fresh reminder of just how quickly sentiment can change: the Philadelphia Semiconductor Index fell more than 5%, even though the sector remained dramatically higher for the year.
So, is this the beginning of an AI bubble—or simply the normal volatility of a genuine technological revolution?
The answer is more complicated than either side suggests.
Table of Contents
Norway’s paradox: record profits and rising fear
Norway’s Government Pension Fund Global is not an ordinary investment portfolio.
With assets around $2.3 trillion, it is the world’s largest sovereign wealth fund and owns roughly 1.5% of publicly traded companies globally. Its enormous portfolio gives investors a unique window into what is happening across financial markets.
And 2026 has been exceptionally profitable so far.
The fund reported a first-half profit of approximately $184 billion, its strongest first-half result on record. Technology stocks, particularly in Asia, played a major role in the gains.
Yet Tangen is not celebrating without qualification.
The problem is concentration.
Around 20% of the fund’s value is now concentrated in its ten largest holdings, many of which are major technology companies, including Nvidia, Apple, Alphabet, Microsoft and Taiwan Semiconductor.
That creates a strange situation:
The fund can make enormous amounts of money because AI stocks are rising—and simultaneously become more vulnerable if those same stocks fall sharply.
That is the paradox at the heart of today’s AI market.
The real warning isn’t that AI is fake
One of the biggest mistakes investors can make is interpreting a bubble warning as an argument against the technology itself.
That is not necessarily what Tangen, Ray Dalio or the Bank for International Settlements are saying.
AI can be transformative.
It can increase productivity.
It can create entirely new industries.
It can change how companies write software, manufacture products, conduct research and provide services.
But a revolutionary technology does not automatically mean every stock associated with it is fairly priced.
That distinction is crucial.
The internet changed the world.
But many internet stocks were still wildly overpriced in 1999.
Railways transformed transportation.
Yet railway investment booms in the 19th century still produced spectacular financial busts.
The same principle could apply to AI.
The technology can be real while the prices become unrealistic.
Ray Dalio sees echoes of 1929 and 2000
Billionaire investor Ray Dalio has raised a similar concern.
Dalio recently argued that the AI market displays characteristics associated with previous speculative episodes and compared today’s environment with the periods surrounding the 1929 crash and the dot-com collapse of 2000.
His argument is less about whether AI works and more about what investors are willing to pay for the promise.
That distinction is easy to miss.
Imagine a company develops a revolutionary technology capable of transforming the global economy.
That may be excellent news.
But if investors bid the company’s stock price to a level requiring decades of extraordinary growth, even a successful business can become a disappointing investment.
A great company can be a terrible investment at the wrong price.
That may be the most important lesson behind the current AI debate.
The CAPE ratio is flashing yellow
Another reason investors are nervous is valuation.
The Shiller CAPE ratio—which compares stock prices with average inflation-adjusted earnings over a decade—has climbed to around 41 in 2026, according to BlackRock Investment Institute data.
That is close to the extreme levels reached during the dot-com era.
But there is an important nuance.
The CAPE ratio does not prove that a crash is coming.
Markets can remain expensive for years.
And today’s largest technology companies are very different from many of the speculative companies that dominated the dot-com bubble. Many AI leaders have enormous revenues, substantial profits and genuine technological advantages.
That makes the comparison with 2000 useful—but imperfect.
The question is therefore not:
“Is AI another dot-com bubble?”
A better question is:
“How much future AI success is already priced into today’s stocks?”
That is much harder to answer.
The trillion-dollar AI spending problem
Perhaps the biggest risk isn’t simply stock valuations.
It is the amount of money being poured into the physical infrastructure needed to build the AI economy.
The Bank for International Settlements has warned that the current AI build-out is among the largest technology-driven investment booms in U.S. history.
Its research estimates that capital expenditure by major hyperscalers could exceed $700 billion in 2026 alone, while aggregate AI infrastructure investment could eventually reach trillions of dollars.
That creates a difficult economic equation.
Companies are spending enormous amounts today based on the expectation that future AI demand will justify those investments.
But what if demand grows more slowly than expected?
What if AI services become cheaper?
What if competition drives margins down?
What if companies discover that some AI projects produce impressive demonstrations but disappointing financial returns?
Then the problem isn’t simply that one stock falls.
The entire investment chain could slow down.
Data centers could be delayed.
Chip orders could be reduced.
Equipment manufacturers could see weaker demand.
Energy projects could be postponed.
Debt used to finance expansion could become more expensive to service.
And investors could begin questioning the assumptions behind the entire AI spending cycle.
Why today’s chip-stock volatility matters
Tuesday’s market action offered a small example of that sensitivity.
The Philadelphia Semiconductor Index dropped more than 5% on August 18, with several chip companies suffering significantly larger losses. Yet the semiconductor sector remained dramatically higher year-to-date after its earlier surge.
That combination is important.
It shows that investors can simultaneously believe in the long-term AI story and aggressively reduce risk when valuations or macroeconomic conditions become uncomfortable.
A one-day decline does not signal the end of the AI boom.
But it demonstrates how quickly a crowded trade can move.
The BIS has a different reason to worry
The Bank for International Settlements—the institution often described as the central bank for central banks—has also warned about the sustainability of the AI investment boom.
Its research argues that technology investment booms can become vulnerable when companies compete aggressively to capture a potentially enormous future market.
Companies may invest more than the eventual economic returns justify simply because they fear falling behind competitors.
This creates a dangerous feedback loop:
AI optimism → higher stock prices → easier financing → more AI investment → stronger AI expectations → even higher valuations.
The problem begins when that loop reverses.
Lower expectations → falling stock prices → tighter financing → reduced investment → weaker demand → even lower expectations.
That is how an investment boom can turn into an investment bust.
The BIS has specifically warned that a reversal in AI optimism could have broader financial consequences because of rising leverage and the expanding role of AI-related companies in credit markets.
But there’s a crucial reason this isn’t 2000
There is another side to the story.
The AI boom isn’t built entirely on companies with no revenue and no customers.
Today’s leading AI beneficiaries include some of the world’s most profitable businesses.
They are generating real cash flow.
They are selling real products.
And businesses are already using AI for coding, customer service, data analysis, research and automation.
The BIS itself acknowledges that the productivity potential of AI could be significant, even while warning that the timing and scale of the payoff remain uncertain.
That makes today’s market fundamentally different from the most extreme examples of the dot-com era.
The technology may ultimately justify enormous investment.
The uncertainty is how much investors should pay today for profits that may arrive years from now.
What could actually trigger an AI correction?
A major AI selloff wouldn’t necessarily require AI to fail.
Several less dramatic events could trigger one.
1. AI earnings disappoint
If the world’s biggest technology companies continue spending hundreds of billions on AI infrastructure but cannot demonstrate comparable revenue growth, investors could begin questioning their returns on investment.
2. Interest rates stay higher
AI stocks tend to depend heavily on expectations about future growth. Higher interest rates can reduce the present value investors assign to those future profits.
3. AI competition destroys margins
More capable models could eventually become cheaper and easier to access.
That would be fantastic for consumers.
But it could make it harder for companies to maintain extraordinary profit margins.
4. Infrastructure becomes excessive
If companies build data centers faster than actual AI demand requires, capacity could eventually exceed demand.
That would put pressure on returns.
5. Geopolitical shocks hit the semiconductor supply chain
The AI ecosystem depends on a highly interconnected global semiconductor industry.
Any major disruption involving trade, Taiwan, energy or other geopolitical flashpoints could quickly affect the entire technology supply chain.
6. Investors simply decide the prices have gone too far
Perhaps the simplest trigger is also the most powerful.
Markets don’t need a catastrophe to correct.
Sometimes investors simply decide that future growth is already reflected in the stock price.
What Norway’s warning really tells investors
The most interesting part of Norway’s warning is not that the country’s wealth fund expects an AI crash tomorrow.
It doesn’t.
The message is about risk concentration.
Norway has built one of the world’s most diversified investment portfolios.
Yet even that portfolio cannot completely escape the consequences of a dramatic decline in the largest technology companies because those companies have become such a large part of global indexes.
That is an important lesson for ordinary investors.
Diversification by owning several technology stocks may not provide as much protection as it appears if those companies all depend on the same underlying AI investment cycle.
Nvidia, chip manufacturers, cloud providers, data-center operators and AI software companies may look like different investments.
During a major AI correction, however, investors could discover that they are connected parts of the same trade.
So, is an AI stock bubble about to burst?
Nobody knows—and that uncertainty is precisely the point.
The evidence does not establish that a crash is imminent.
Instead, it shows that the market has entered a period where expectations, valuations, capital spending and concentration have become unusually important.
Norway’s sovereign wealth fund is warning.
Ray Dalio is warning.
The BIS is warning.
And today’s semiconductor selloff demonstrates that AI-related stocks can still experience violent moves even while the long-term technology story remains intact.
But there is another possibility that is often ignored.
AI could genuinely transform the global economy—and investors could still lose money by paying too much for that transformation.
That is the paradox.
The technology doesn’t have to fail for the bubble to deflate.
It only has to deliver less growth than investors have already priced in.
And that may be the question Wall Street will be asking for years:
Will AI grow fast enough to justify the enormous expectations already built into its stock prices?
For now, Norway’s message is less a prediction of an imminent crash than a warning not to confuse technological progress with guaranteed investment returns.
The AI revolution may be real.
The price investors are paying for it is the part that remains uncertain.
What to watch next
Investors looking for signs that the AI trade is becoming healthier—or more dangerous—should watch five things closely:
- AI revenue growth versus AI infrastructure spending
- Data-center returns and utilization
- Profit margins at major technology companies
- Semiconductor demand and inventory levels
- The valuation gap between AI leaders and the broader market
If AI earnings continue accelerating fast enough to justify today’s spending and valuations, today’s bubble warnings could eventually look premature.
If spending keeps rising while returns disappoint, however, the market may discover that the most expensive part of the AI revolution wasn’t building the machines.
It was the expectations investors built around them.
Recommended for you:
- The $100 Million Secret Nobody’s Talking About: Why Your Company’s Slack Messages Are Now Worth More Than Your Product
- The $60 Billion Question Nobody’s Asking: Did SpaceX Just Buy Cursor to Train Grok, or Did Cursor Just Take Over SpaceX?
- The Real Reason Intel’s $20B Stock Offering Should Worry the Entire Semiconductor Industry