Warren Buffett, the Oracle of Omaha, once said:
Never ask a barber whether you need a haircut.
An irreverent remark that conceals a very simple and, in fact, highly rational principle. Anyone with a direct interest in an answer is not the right person to ask for an opinion.
A barber will always need to cut hair. In the same way, the analyst running a fund exposed to Nvidia will always, always need Nvidia to go up.
Right now, almost every source of stock market advice has a direct interest in how those markets perform. And almost nobody says so.
Consensus is not analysis
Seventy institutions, seventy variations on the same message.
To be clear, this does not mean they are wrong. But it may also mean that the consensus is the product of a system in which the institutions producing the forecasts are the same ones that manage the funds, run the placements and earn the transaction fees.
The barber is not lying when he says you need a haircut. He genuinely believes it. But his point of view is skewed.
Nvidia: the most analysed stock in the world
Revenue for the third quarter of fiscal 2026 was $57 billion, up 62% year on year. CEO Jensen Huang described demand as “insatiable”.
Nvidia’s price-to-earnings ratio currently stands at around 41 times forward earnings, against a three-year average of 61 times. A multiple of 41 is not necessarily irrational for a company growing at 62%, but it builds in the assumption that this growth continues, that demand for AI GPUs does not normalise, and that no competitor manages to erode the 92% data-centre market share Nvidia holds today.
The main risk is multiple compression: if growth slows even marginally against expectations, the market will correct sharply, because high multiples do not forgive downside surprises.
Whoever advises buying Nvidia today almost always already holds Nvidia in their portfolio. That does not make the advice wrong. But it certainly does not make it independent either.
SpaceX: an act of faith
For context: Saudi Aramco, the largest IPO in history by market capitalisation at debut, reached 1.7 trillion. SpaceX would aim to surpass it.
The problem lies in the numbers. In the twelve months before the merger with xAI, SpaceX had posted revenue of around 15-16 billion with around 8 billion in operating profit. After the merger, 2025 closed with an operating loss of $4.9 billion: xAI burns cash on a massive scale, and the combined cash flows are what the more cautious analysts describe as “profound” in their unpredictability. At a valuation of 1.75 trillion, the price-to-sales ratio would be 91 times revenue over the last four quarters. The price-to-book ratio would be 50 times, excluding the cash raised in the IPO.
AkademikerPension, a $25 billion Danish pension fund, has blacklisted SpaceX, describing its governance structure as “catastrophic”. Elon Musk is simultaneously CEO, CTO and chairman of the board, whose very mandate is to oversee management. The fund stated that even at an acceptable valuation, it would not have invested on governance grounds.
Those pushing for the SpaceX IPO are the private-round investors who need an exit. The investment banks managing the offering earn fees in proportion to the amount raised. The once-in-a-lifetime-opportunity narrative is produced by those with a direct interest in the deal’s success.
The barber, again.
The neuroscience of roller coasters
The dopaminergic system responds above all to the anticipation of reward.
When the market rises, when stories of extraordinary gains circulate and the consensus is optimistic, the brain interprets exclusion as loss. Status quo bias is reversed: staying out becomes the choice perceived as riskier, not getting in.
Research on confirmation bias shows that under high emotional arousal the brain tends to seek out information confirming the direction already taken or desired, and to discount information that contradicts it. In a market where every financial publication, every investment newsletter and podcast repeats the same message, the cognitive cost of dissent becomes very high. And this is because the information system has been built to make consensus comfortable and dissent uncomfortable.
The precedent
In 2000, on the eve of the dot-com crash, the Wall Street consensus was just as universal. The internet was the greatest economic transformation in history. Valuations were justified by future growth. Anyone who stayed out was missing a once-in-a-lifetime opportunity.
Some of those forecasts were correct: the internet did indeed transform the global economy. But between a correct technological forecast and a correct market valuation lies a gap that the Nasdaq took fifteen years to close after 2000.
AI could be the most profound transformation since the Industrial Revolution. Nvidia could still be worth 5 trillion ten years from now. SpaceX could become the largest company in the world. All of this could be true at the same time as current valuations already pricing in much of that future growth, and sometimes more.
The question every investor should be asking, then, is not whether AI will change the world (it will). The question is whether the price they are paying today reflects what they already know, or what they hope.
How to answer that question
So let us focus on the real message of this article, and of the Oracle’s remark, which holds true today more than ever.
Warren Buffett never said not to get your hair cut. He simply observed that asking a barber for an opinion on the matter does not always lead to the most objective answer.
Conflicted sources do not necessarily produce bad advice. That advice can sometimes be right, but always incidentally, because the priority lay in self-interest. Not yours: theirs.
In a market dominated by narrative enthusiasm, finding sources whose primary aim is your interest is difficult, but necessary.
At this delicate stage for the markets, then, with SpaceX’s IPO and a major upheaval in overall share prices on the horizon, carefully analysing the sources of financial guidance matters more than anything else.
Because the consequences could be far worse than a haircut you didn’t need.
New Connections (FAQ)
But if seventy financial institutions are optimistic about AI, aren’t they likely to be right?
They may be right about the direction and wrong about the valuation. The two are not mutually exclusive. In 2000 the consensus was correct that the internet would transform the economy. It was wrong about the prices at which it was worth buying. Seventy institutions in agreement do not produce a more accurate analysis than one: they produce seventy analyses built under structurally similar conditions, with similar incentives, for clients with similar expectations. The correlation of consensus is no proof that it is correct.
Nvidia is growing at 62%. Isn’t that enough to justify the valuation?
The market does not pay for past growth. It pays for expected future growth. A multiple of 41 times earnings builds in the expectation that this growth will continue. If it slows from 62% to 30%, Nvidia could still be a great company and lose 40% of its market value in the same quarter. The relevant question is not whether Nvidia is an excellent company. It is whether the current price is right for an excellent company with that specific growth trajectory.
Should I avoid investing in AI altogether?
That is not the conclusion of this article. The question is not whether to invest in AI. It is at what price, over what time horizon, and on the basis of what independent analysis. There are significant differences between buying Nvidia at 41 times earnings, buying a diversified AI ETF, investing in companies that use AI to improve their margins at reasonable valuations, or waiting for a correction to enter at prices that reflect less euphoria. These are different choices with very different risk profiles.
How do I find genuinely independent analysis?
By looking for those who do not profit from your investment. Investment bank reports are written by those who manage offerings. Paid newsletters are a step forward, but some depend on retaining readers who are already in the market. Short-seller funds publish critical analyses with an interest opposite to the consensus: they are not neutral, but they flag risks the market tends to ignore. Academic research is slower but structurally more independent. The simplest indicator: if the person advising you profits when you follow their advice, they are not your most reliable source.
N.B. The following article is for information purposes only. The information contained herein does not in any way constitute financial advice. Before investing, we encourage you to do your own research (DYOR) or to consult a certified financial adviser.