Nicolai Tangen is the man behind Norway's sovereign wealth fund, a kind of mega fund owned by the small but oil ultra rich country and which owns approx 1,5% of every publicly listed company on this planet. To put this in perspective, we talk about approximately 7.200 leading companies across 60 countries.
When a guy like this wakes up in any given morning, before he had his first coffee, the needle of his 2,1 trillion might be moving up or down by a number of geopolitical tensions, currency fluctuations, interest rate decisions or earning surprises in industries he doesn't even particularly follow.
His job at the wheel of this giant is basically to see things coming before they arrive, and calculate the outcomes of any chosen investment or industry the fund is invested in, so when Tangen jumps into an interview saying that an AI bubble is one of the two greatest threats to global markets right now (the other according to him is geopolitical fragmentation), it is worth taking seriously.
About a month and a half ago, his fund formally identified that a potential AI bubble would be a major risk scenario that could potentially erase 35% off the fund's value in a worst scenario case, an along interviews where he participated between last year and this one, Tangen has been consistently saying that the concentration of capital into AI related stocks is creating, in his own words, "a risk we have never seen before".
He thinks that basically the market may be running ahead of what the technology can actually deliver in the near term, and by checking his funds investments anyone can see that he is definitely not a technophobe making this argument from ignorance. In the contrary, he has spent years (in his own words) "running around like a maniac" to get his 700 staff to use AI in their daily work.
His fund actually right now uses large language models, including Claude, to analyse every new equity investment on the day it enters the portfolio, generating daily risk assessments in real time, and a very big number of his employees code their own AI tools. So Tangen is a genuine believer in what the technology does, but his concern goes way further and is about what the market is actually pricing in. That differentiation (between the technology and the valuation), is for him the key of everything.
If we are to give a true analysis, the concentration argument is quite hard to dismiss when we see things like seven single companies now representing roughly a third of the entire SP500. Another indicator is the Shiller CAPE ratio, a inflation oriented measure that has historically preceded major corrections and that now sits around 38 to 40, levels only exceeded at the dotcom peak of 44.
We can clearly see that the market is pricing in a future where AI delivers transformational returns at a speed and scale that, looking at enterprise adoption data, is not yet materializing.
And maybe the part of Tangen's argument that is most interesting is what he describes as a structural trap, with him explicitly saying that even if he believed AI valuations were stretched, he could not simply sell out of the AI companies because they are already too large a share of the market that for a fund his size, reducing exposure to giants like Nvidia, Microsoft or Apple would mean moving markets against himself (basically selling into a decline he helped starting)
He is, according to his own words, an investor with "skin in the game" in a way that it even affects his own investing behavior. That is an uncommon and honest admission from someone in his position, and it also points to a systemic fragility that most optimistic arguments quietly diminished.
The enterprise adoption data gives Tangen's caution some realistic sense too, as despite extraordinary capital investment, most of the productivity gains from AI are still showing up in individual measurements rather than in corporate earnings at the scale the current impressive valuations imply. The MIT study finding that 95% of enterprise AI pilots fail to scale is not a detail that trillion dollar valuations should be able to ignore indefinitely.
In any case, there are also points on Tangen's speech that might be incorrect or at least incomplete, because there is a problem with the bubble analogy that we already explored in a previous article a few weeks ago: The companies at the center of the AI rally cannot be compared with those of the Dotcom era, simply because they are generating cash at a scale that has no historical precedent, not even close.
Well known companies like Apple, Microsoft, Alphabet, Amazon and Meta combined for roughly 350 billion USD in free cash flow in their most recent fiscal years, with Nvidia alone marking a net income exceeding 120 billion USD in 2026. These are not speculative bets on future revenue that may never materialize, as it clearly was in the case of previous tech crashes. They are simple audited financial statements from the most profitable enterprises in the history of capitalism.
But the infrastructure argument is also something that the bubble narrative struggles to fully explain, when we see companies like Microsoft independently committing more than 80 billion USD to AI infrastructure, Alphabet 85 and Meta between 115 and 135 billion USD. These are accounting decisions funded by existing free cash flow, not leveraged bets financed by faithful or blind investors.
The companies building AI capacity and infrastructure have among the strongest balance sheets in the entire equity market, and are setting their decisions not by guessing but by building with a financial cushion that no Dotcom company ever had.
And finally there is the adoption curve itself, a factor where Tangen is also conservative but where we should be confortably positive about when we see that enterprise AI deployment is still, as of now in mid 2026, in measurable very early stages, with companies like Gartner projecting that 40% of enterprise applications will integrate AI agents by end of 2026, up from less than 5% in 2025. If that curve materializes, even if it's on a 50% basis of what Gartner expects, the earnings growth that would justify current valuations hasn't happened yet. It's just coming.
In any case, what makes Tangen a more interesting voice than most market commentators (apart from all his professional validations) is that he doesn't claim certainty. Actually, when he presented his fund's results back in early 2026, he said openly that it was "very difficult to gauge right now whether there was an AI bubble". So this means he is basically not anticipating the crash but instead he is flagging the risk, managing the uncertainty and being open about the limits of his own visibility. So his vision as a whole might make sense and is actually the correct response to the situation, leaving aside other superbullish or crash anticipating arguments.
The most clear version of Tangen's argument is not just a vague "AI is a bubble", but he is more meaning that the concentration of capital in AI related companies, combined with a mismatch between current valuations and earnings realization, creates a correction risk that is larger than most people are pricing in. That is a serious and defensible position that doesn't come as a prediction but as a risk assessment from someone who, by the nature of his job, is paid to think about risk.
And to defend his opinion, a 30-35% correction in AI related equities would not be crazy because the history of transformative technologies is full with periods of genuine overvaluation followed by painful corrections followed, eventually, by the technology delivering on most of its original promise.
Internet did change everything but when we look back we realise that it took a decade longer than 1999 suggested it would, and most of the companies that were going to lead that change didn't exist yet when the bubble exploted.
AI probably will change everything too. Whether the companies currently priced to do so will be the ones that deliver it, and whether that delivery will happen on the timeline the market is currently pricing is a question that even the man managinf one of the largest funds in the world says he really cannot answer, which means the rest of us should probably be thinking about that too.
Sources:
AI bubble as a risk scenario (35% fund loss)
Bloomberg, March 2026:
https://www.bloomberg.com/news/articles/2026-03-18/norway-s-wealth-fund-warns-of-ai-bubble-and-geopolitical-risks
Bloomberg video, April 28, 2026 (markets, AI and China):
https://www.bloomberg.com/news/videos/2026-04-28/norway-s-tangen-on-markets-real-estate-ai-and-china-video
"Managing $2 Trillion: AI Bubbles & Contrarian Investing", February 2026:
https://www.youtube.com/watch?v=zyvuM3J9QqQ
Seven IT giants owning 1 third of the SP500:
https://luisyanguas22.medium.com/seven-it-companies-now-own-a-third-of-the-sp500-heres-what-this-actually-means-for-developers-2803c43cf093
IPE — testimony to Norwegian parliament on "unprecedented concentration risk", May 2026:
https://www.ipe.com/news/nbims-tangen-warns-lawmakers-of-swfs-unprecedented-big-tech-concentration/10136514.article
CNBC — fund using Claude AI to screen investments + AI is "deflationary and positive" quote, April 2026:
https://www.cnbc.com/2026/04/28/norway-sovereign-wealth-fund-oil-iran-war-anthropic-claude-ai-invest-stocks.html
The incredible transformation of modern software by AI:
https://translockit.com/en/article/the-incredible-transformation-of-modern-software-development-by-artificial-intelligence
www.translockit.com
Luis Carlos Yanguas Gómez de la Serna
Top comments (0)