The Simple Test To See Through AI Stock Hype

How to tell how much of an AI stock’s price is real earnings – and how much is just hope.

• A soaring stock price alone doesn’t prove bubble risk – the key is whether earnings are rising fast enough too.
• PVGO helps investors determine how much of a stock’s price is based on today’s proven profits and how much on tomorrow’s hoped-for growth.
• The question is less about whether AI is real, but whether future profits can justify the price investors are paying today.

Right now, a lot of people want to know if AI stocks are in a bubble. And for good reason: valuations are high and just seem to keep getting higher.

But I don’t think that’s the most useful question to ask.

Here’s a better one: how much of the price is backed by what these companies already earn – and how much depends on what investors hope they’ll earn later?

That distinction matters. A stock that doubles isn’t automatically expensive. If earnings quadruple at the same time, it may actually become more grounded, not less. The share price tells you what investors paid. It doesn’t tell you what they’re paying for.

That’s especially important now because the old valuation shortcuts are struggling. Traditional value investing often compares a stock with book value – the accounting value of its assets. That worked better when companies were built around factories, machinery, stores, and inventory. Today’s biggest winners are built on software, chips, brands, networks, patents, data, and know-how – assets that don’t always show up neatly on a balance sheet. And that’s also why classic value investing has struggled since the early 2000s: a low price-to-book stock isn’t automatically cheap and a high one isn’t automatically expensive.

So AI needs a better test. And that means separating what a company already earns from what investors hope it will earn later.

Every stock price is really two prices

A stock price is really two prices stitched together.

• Bird in the hand: The first part is the value of the business as it exists today, its steady-state value. Think of a coffee shop that reliably earns $100,000 a year. If it can keep earning that without opening new stores, launching new products, or doing anything heroic, that regular earning power is the “bird in the hand”.
• Two in the bush: The second part is the value of what the business might become. The coffee shop might open 50 new branches, launch a packaged coffee brand, or build a subscription app. Those opportunities may be valuable, but they are not guaranteed, so they’re the “two in the bush”. Or, in finance lingo, PVGO: the present value of growth opportunities.

So imagine a company is worth $100 billion in the stock market. Its current earnings power might justify $60 billion of that. The remaining $40 billion is the market’s bet on future growth. In that simple example, 40% of the price is PVGO.

That number is not a buy or sell signal. A high-PVGO company can be an exceptional business. It may have a huge market, a strong moat, and lots of ways to reinvest at high returns. But high PVGO means investors are paying upfront for future success. And that’s how bubbles form. The future can be real, but investors can still pay too much for it too early.

That brings us back to AI. It clearly has real economic value – but the hard part is working out whether the profits will be large enough, soon enough, and durable enough to justify today’s prices.

High hopes, thin margin for error

The market-level data shows why this matters now.

Morgan Stanley looked at the S&P 500’s PVGO from 1961 to 2025. Over that period, PVGO averaged 35% of the index’s price, while steady-state value – that bird in the hand – accounted for the remaining 65%.

But by the end of 2025, PVGO was in the mid-50s – well above the long-run average. Put simply, investors were paying more than usual for future growth that still had to be delivered. That doesn’t mean stocks were automatically in a bubble, but it does mean the margin for error was thinner.

The price-to-earnings ratio (P/E) multiple points to the same underlying issue: investors are paying a lot for growth that still has to arrive. Based on an 8.75% cost of equity – the return investors demand for owning stocks – Morgan Stanley estimates that the steady-state value of the US stock market would justify a P/E multiple of 11.4.

But at the time of the analysis, the S&P 500 was trading at 21.9 times expected earnings for 2026, nearly double that estimate. That gap is the market paying extra for future growth or assuming today’s earnings are temporarily depressed, or some combination of both.

High expectations can be justified – but when investors have already paid a lot up front for them, even small disappointments can hit the share price hard. Historically, lower-PVGO markets – less driven by big growth hopes – have tended to produce better future returns. The lowest PVGO quartile was followed by 10-year annualized total shareholder returns of 11.6%. The highest PVGO quartile, however, was followed by 7.6%.

Morgan Stanley is clear that PVGO is not a precise timing tool – it can’t tell you when to buy or sell. The correlation with future 10-year returns is only moderate. Still, the pattern is useful when expectations reach extremes. Traditional valuation labels – low or high P/E multiples – can only tell you how a stock’s price compares with its future earnings. But a lower-multiple AI stock can still be expensive if its current earnings are peaking, while a higher-multiple AI stock can still be reasonable if it is already turning demand into durable profit. The label matters less than the evidence.

This is where PVGO can be more useful than old-school value metrics, especially in an intangible economy built around AI, data, and tech. Book value misses a lot of what makes modern companies valuable. PVGO focuses on expectations, telling you how much of a stock’s price is grounded in today’s earnings and how much is based on what investors hope it’ll earn in the future.

Here’s how you can work it out yourself

You can get a rough sense of PVGO in about a minute, with just two numbers. The first is a company’s price-to-earnings ratio, or P/E. A P/E of 20 means the company’s market value is 20 times what it earns in a year.

The second is a no-growth yardstick. Picture a company that earns the same amount every year, forever, and never grows. Morgan Stanley uses a cost of equity of 8.75%, which implies a steady-state P/E of 11.4. So, as a rough shortcut, treat 11.4 as the price of “just the profits the company already makes”.

Then compare the two. If a stock’s P/E is above 11.4, investors are paying extra for growth that has not happened yet. The bigger the gap, the more the price depends on future success.

The rough formula for what we can call the hope share is (P/E − 11.4) ÷ P/E

A stock on a P/E of 22 has a hope share of about 48%. A stock on a P/E of 14 has a hope share of about 19%. A stock on a P/E of 40 has a hope share of about 72%.

That does not make the P/E 40 stock bad, but it does mean that the company has a lot to prove. Profits need to grow strongly, for years, to justify the price. Meanwhile, a lower-hope stock can surprise investors more easily. If a company priced for almost no growth grows even a little, that upside was barely in the price.

This shortcut is not perfect. It works best for profitable companies with earnings that are not wildly cyclical or temporarily inflated. It also ignores debt and differences in business risk. But it is useful because it turns the bubble debate into a better exercise: working out how much hope is in the price, then deciding whether the company can deliver more than the market already expects.

What the hope has to deliver

The hope share is still only the starting point. The useful move is to translate it into a growth hurdle: how much profits need to rise for today’s price to be backed by earnings rather than hope.

The simple version is the grow-into-it test. It asks how much earnings would need to increase for today’s share price to equal the 11.4x no-growth multiple.

Required earnings increase = current P/E / 11.4 – 1

Annual growth needed = (current P/E / 11.4)^(1 / years) – 1

Take the P/E of 22 example. The stock is trading at almost twice the no-growth yardstick: 22 divided by 11.4 equals 1.93. That means earnings would need to rise about 93% for today’s price to be justified by no-growth earnings power alone.

That same 93% profit increase can arrive quickly or slowly. The shorter the growth runway, the faster earnings have to compound.

The chart shows the annual earnings growth needed for a stock on 22x earnings to justify today’s price under an 11.4x no-growth yardstick, depending on how long the growth runway lasts. Source: Finimize.

The point is not precision. The point is discipline. Once you know how much hope is in the price, you can judge whether the growth required looks realistic, stretched, or plainly heroic.

Ultimately, the AI story is about sorting the winners from the wannabes. The winners will be the companies that turn AI demand into durable earnings. The disappointments will be the ones where investors paid for too much success before it showed up.

—

Originally Posted June 24, 2026 – The Simple Test To See Through AI Stock Hype