The Belief Quotient

I was 24 when a group of angel investors headed by Darren Mercer put a million pounds into a company of mine that had no product and no revenue. What it had was a belief, which I held and they bought, that the internet would need security. This was when the encryption in a browser was an optional extra and the internet bubble was gaining momentum, and the belief was correct and too soon. People were still trying to get things to work at all. A conversation about securing data and applications was a solution to a problem they did not yet have, and the peace of mind they were prepared to pay for came from buying a firewall, even if the firewall secured neither. The company never achieved significant revenue. Its valuation kept rising anyway, carried on the tide of the bubble. In that company the belief was everything there was.

Years later, at CoreAVI, I had the opposite problem. The company had customers and revenue, and it also carried a great deal of inventory, because an aircraft programme runs for 20 years and the parts to support it have to be bought while they can still be bought. Bank after bank refused to finance that inventory, because on the bank’s spreadsheet it was slow-moving stock. To me it was the asset that secured the company’s future. To the bank it was a burden that sank the balance sheet. The first company had a story and no numbers, and the market priced the story at a million pounds and rising. The second had numbers and a story the bank would not read, and the market priced that story at less than nothing. I have spent 30 years between those two seats, and it is why the chapter of my book that people keep asking me about is the one on what belief costs.

When I told friends and family that I was writing a book called Disbelief, almost every one of them assumed it would be about religion. An atheist writes a book on belief; where else would he go. The book does discuss religion at some length, and I am content with what is in it. The chapters that have paid back most reliably, more than a year after it went to press, are not the ones that examined religion, though. They are the ones that examined belief in places most readers would not have expected to find it: markets, branding, politics, cognition. The most operationally consequential belief system in the life of a modern professional is, in almost all cases, not the religious one.

The chapter I keep being asked about, by investors and by people running companies of their own, is the one I called the Belief Quotient. In any public-company valuation, a portion is explained by financial fundamentals: revenue, earnings, growth, cash on the balance sheet, the rate at which the company is winning or losing against its competitors. The rest is what the market is willing to pay above what the spreadsheet says. I called that residual the Belief Quotient, because it is in every meaningful sense a belief, made tradable and priced. A year on, the gap is larger than when I wrote the chapter, and the largest companies on the public markets in 2026 are priced on belief much more than on history. I have spent 30 years as a founder talking to investors who were pricing my own ventures on the same basis, and this is the version of the argument that operates from inside that seat.

What the Belief Quotient Is

The price-to-earnings ratio is the cleanest way into the calculation. A price-to-earnings ratio, the multiple I use throughout this essay, divides what a company is worth, by share price, by the earnings it produces. A typical mature business trades at a multiple between 13 and 15; an investor buying at that level is accepting an earnings yield of around 7 per cent and assuming the business will continue to perform roughly as it has done. That mature multiple is the yardstick I use for everything that follows. Take it as 15, divide it by the multiple a company actually trades at, and the result is the share of the price that historical earnings can account for; the rest is the Belief Quotient. A company at 30 is half belief. A company at 150 is 90 per cent belief. A company at 1,000 is belief with a rounding error of history attached. Companies expected to grow faster than the market trade at higher multiples, 30 to 50, sometimes more, and the buyer is paying for growth that has not yet happened.

The interesting cases are the companies that trade at multiples no plausible forecast can justify. Tesla in 2021 is the cleanest example. The company had become one of the ten largest businesses on the planet by market capitalisation, passing a trillion dollars in October of that year, while holding little more than 1% of the global car market and around 14% of the electric-vehicle segment it had created. Its multiple had crossed 1,000 in January. To bring it into line with the other companies of its size, Tesla would have had to grow its earnings thirtyfold in a market that was visibly becoming more competitive. No spreadsheet, however generous, produced that result. Investors were paying for something else.

Apple’s history locates the size of the gap. At the launch of the iPhone in January 2007, Apple’s peak multiple was 41, against a historical average of 22: on the yardstick, about a third of its valuation at that moment could be accounted for by historical earnings, and the other two-thirds was the belief that the new product would compound. Amazon takes the comparison further. Between 2012 and 2013 its multiple crossed 1,000 at a valuation of about $100 billion, with the market assuming the company would grow into the cloud-services and logistics business it eventually became. Roughly 98 per cent of Amazon’s valuation at that point was the belief and the rest its history, and the market was substantially right; Amazon grew ten times larger over the next decade. For Tesla in the autumn of 2021, at a multiple in the hundreds, around 5 per cent of the stock price could be attributed to historical financial performance, and in January, above 1,000, less than 2 per cent. The remainder was Belief Quotient. This is not a metaphor. It is the proportion of the price being paid for what investors thought would happen rather than for what the company had so far done.

The objection from the efficient-markets side is that none of this is belief. The residual above the mature multiple is expected growth, priced by people with more information and more at stake than I have, and calling it belief is a rhetorical move that lets an outsider feel superior to the market. I accept the first half. The residual is expected growth. My claim is that expected growth ten years out is a belief by any definition that matters: held with more or less evidence, more or less contagious, and repriced when a story breaks rather than when a number changes. The day I come to next is the demonstration. If the market had been pricing a forecast, Nvidia’s fall would have been proportionate to a revision in the forecast. It was proportionate to the collapse of a story, and it came back when the story was rebuilt.

The Charismatic Catalyst

The Belief Quotient is highest when there is a person to believe in. Belief in a company can attach to a category, a product line, a regulatory moat or a balance sheet, but it compounds fastest, and reaches the highest valuations, when it attaches to a particular person, because a person can carry conviction in a way that a balance sheet cannot. Tesla’s valuation in 2021 was not separable from Elon Musk’s personal brand. Apple at the iPhone moment was Steve Jobs in the same sense. Nvidia, in 2025 and 2026, is Jensen Huang in the same sense. The Belief Quotient in these cases is not assigned to a corporation. It is assigned to the conviction that this particular person will deliver the future the market is pricing in.

Jeff Bezos at Amazon is the partial counter-example that makes the rule visible. Amazon’s Belief Quotient was real in 2012 and 2013, but it never had the devotional quality that Tesla and Nvidia have had. Bezos was unusually competent rather than charismatic in a way the market could index, so Amazon’s belief attached to the business model, not to the founder, and when he stepped back from the chief executive’s role the share price did not collapse. It moved on. Charismatic-leader belief is leverage in both directions. The chief executive who can carry the narrative is a multiplier on the company’s valuation, and the one who stops carrying it is a multiplier the other way. The personality that lifted the Belief Quotient becomes the single point at which it can be broken.

The DeepSeek Moment

The cleanest illustration of how quickly a Belief Quotient can compress is dated, exactly, to 27 January 2025. On that day a Chinese artificial-intelligence firm called DeepSeek, which two weeks earlier had been unknown to almost every investor in the world, released a model that performed competitively against the leading systems from OpenAI, Google and Anthropic at a fraction of the computing cost. Nvidia, whose chips had been the dominant input to the existing high-cost training runs, lost $588.8 billion of market capitalisation in a single trading day, the largest single-day loss of value for any company in the history of public markets.

Nvidia’s chips were the same chips on the 28th as they had been on the 26th. Its revenue, margins, balance sheet and customer relationships were the same. What changed was that the market had been pricing in a particular shape of the AI economy, extremely high training costs, a small number of buyers who could afford them, and a long-lived moat around the leading models, and DeepSeek demonstrated that the moat was thinner than assumed. The belief did not survive the demonstration. The price did not survive the belief.

This is the operational signature of a high Belief Quotient. Companies whose valuation is composed mostly of historical earnings depreciate gently. Companies whose valuation is composed mostly of belief do not depreciate at all until they collapse. The revaluation, when it comes, moves with the speed of a news cycle, not a P&L, and there is no halfway state. The story holds, and then it does not. I am not making a prediction about Nvidia, which by mid-2026 had recovered most of the lost capitalisation and was trading at a multiple of around 45. I am making a structural observation. Belief Quotients are exposed to information shocks in a way that earnings multiples are not, and the shocks do not arrive on a schedule.

The Operator’s Three Seats

If you work in or around public markets you occupy more than one seat in relation to the Belief Quotient, and the discipline is different in each. Most professionals I have known never consciously separate them.

The first seat is the producer. As a founder or an executive responsible for a public-facing narrative, you are in the business of generating belief alongside running operations. The market is paying for the story as much as for the spreadsheet, and pretending otherwise will, over the cycle, cost your shareholders money. Producing belief is not the same as deceiving the market. It is the discipline of articulating why the future is going to be different from the past in a way the market can price. I have spent 30 years in this seat, and the version of the discipline I trust is that you have to hold two things at once: the unreasonable confidence that the narrative will deliver, and the brutal honesty about whether the underlying business can survive if it does not. It is the same gate I described in The Entrepreneur’s Burden. The producer who has only the unreasonable confidence is constructing a route to ruin for everyone holding the stock.

The second seat is the consumer. As an investor you are buying belief on behalf of capital that does not belong to you, or that does but cannot afford to be wrong. The classic statement of this seat’s discipline is Warren Buffett’s: be fearful when others are greedy, and greedy when others are fearful. That is, at root, a statement about the Belief Quotient. The herd pays the most when the quotient is highest, which is also when its compression risk is highest, and pays the least when the quotient is being repriced downward, which is when the underlying earnings have, by definition, become better value. The consumer’s edge depends almost entirely on reading the gap between price and fundamentals and acting against the prevailing belief. Buffett’s preferred holding period, he has said many times, is forever, which implies that he expects belief to wash up and down a great deal and intends to ignore it.

The third seat, and the one most professional readers underestimate, is the hostage. If you are an executive or employee whose compensation includes meaningful equity, you cannot sell when you want to, often cannot sell for years at a time, and the value of what you have been promised is, in the bulk, Belief Quotient rather than historical performance. The grant looks superficially like the producer’s exposure. It is structurally closer to the consumer’s, but without the consumer’s ability to act. The right description is hostage, and most professional readers I have put this framing to find it uncomfortable and accurate.

The producer’s job is to generate belief and stay solvent. The consumer’s job is to buy belief at a discount and sell it at a premium. The hostage’s job is to recognise that they are the one paying for the producer’s narrative, and to diversify accordingly. Confusing the seats is one of the more reliable ways to lose money in public markets.

What the Market Is Charging

The reason this chapter of Disbelief has held up better than the religious chapters, I now think, is that financial belief is the form of belief the modern professional encounters most often, has the most invested in, and reflects on the least. Market valuations get treated as outputs of analysis when they are mostly outputs of conviction, and charismatic founders as outputs of their balance sheets when their balance sheets are mostly outputs of them. I have treated my own equity compensation as deferred salary when it is mostly a leveraged position on a belief whose price I do not control.

The AI cycle has produced, by every measure I can find, the largest concentration of high-Belief-Quotient valuations in the history of public markets. Nvidia is past four trillion dollars. Palantir trades at a multiple above 150. A long tail of smaller AI-adjacent companies trades at multiples that only make sense if you accept that the belief is correctly priced. Some of it is. Some of it is not. Forecasting which is which is not the discipline I am recommending. The discipline I am recommending is to know which seat you are in, what the seat asks, and what its mistakes look like.

At 24 I was paid a million pounds for a belief. At CoreAVI I could not borrow against a warehouse of parts, because nobody would price the belief that a 20-year programme would run for 20 years. I have been on the receiving end of the quotient and on the paying end, and the only discipline I trust is to know, in any given transaction, which end I am on.