Nvidia has become so valuable that almost any discussion of its stock eventually sounds absurd.

The company is worth roughly $5.6 trillion.

Its shares have climbed dramatically during the AI boom.

Its chips sit at the center of the world's most aggressive data-center spending cycle.

And yet one of Asia's prominent investment strategists is making an argument that sounds almost heretical after two years of AI mania:

Nvidia does not look like a classic bubble stock.

DBS Group Chief Investment Officer Hou Wey Fook points to a surprisingly modest valuation relative to the company's expected earnings growth. Nvidia is trading at roughly 17 times 12-month forward earnings, while analysts expect earnings growth of about 70% next year, according to Bloomberg data cited by Yahoo Finance. Hou contrasted that with Cisco's valuation before the dot-com crash, when the company traded near 100 times earnings.

His argument is simple.

A stock can be gigantic and still not be a bubble if earnings are growing fast enough to justify the price.

That does not mean Nvidia is risk-free.

It means the AI valuation debate is more complicated than a comparison with the dot-com era suggests.

Market capitalization is emotionally powerful.

When Nvidia became the world's most valuable publicly traded company, the headline number seemed almost disconnected from economic reality.

But market cap alone tells investors very little about valuation.

A company worth $5 trillion can be expensive.

It can also be cheap.

The real question is how much profit that company is generating and how quickly those profits are expected to grow.

That is where Nvidia's current setup becomes unusual.

The company is not merely being priced on the promise of eventual earnings.

Its existing profits have already exploded.

Nvidia's second-quarter fiscal 2027 revenue reached $96.2 billion, up 106% year over year, while data-center revenue jumped 117% to $89.0 billion.

Those are not startup numbers.

They are the numbers of a technology infrastructure giant operating at historic scale.

This is the central argument behind the bullish valuation case.

When a company's stock rises rapidly, the price-to-earnings multiple can remain stable—or even fall—if earnings rise faster than the stock price.

That is exactly what has happened with Nvidia.

Reuters reported in September that Nvidia's forward P/E had fallen to about 16.5 times, its lowest level since January 2015, despite the company's extraordinary expansion. Its 15-year average was around 30 times.

The ratio has fallen partly because analysts have dramatically increased their earnings expectations.

That is a very different situation from a stock whose price surges while profits remain speculative.

In a classic bubble, investors often pay enormous multiples for earnings that may arrive years later—or never.

Nvidia is producing massive profits now.

The debate is whether today's extraordinary profits can continue.

That is a harder and more legitimate question.

One of the biggest catalysts for the renewed valuation argument came in August.

Nvidia provided an unusually ambitious long-term forecast, projecting roughly 70% revenue growth for its next fiscal year and signaling a potentially much larger AI-computing opportunity through 2028. The announcement helped trigger a broad rally across semiconductor stocks because it reduced fears that AI infrastructure spending was about to slow sharply.

That guidance matters because AI investors have spent much of 2026 wrestling with one big question:

How long can the spending boom last?

If hyperscalers and AI laboratories suddenly reduce capital expenditures, Nvidia's earnings growth would eventually feel the impact.

If spending continues accelerating, today's valuation could become less demanding as future earnings arrive.

The entire investment case therefore depends on the durability of AI demand.

Nvidia has also done something very unusual for a growth company operating at the center of an aggressive investment cycle.

It has decided to buy back an extraordinary amount of its own stock.

In September, Nvidia increased its buyback authorization by $150 billion, bringing total remaining repurchase capacity to approximately $235 billion through fiscal 2028. The increase surpassed Apple's previous record authorization.

Why does that matter?

Because buybacks can increase earnings per share by reducing the number of shares outstanding.

It also signals that Nvidia's management believes its enormous cash generation gives it room to invest heavily while still returning capital to shareholders.

The company has not chosen between growth and capital returns.

It is attempting to do both.

That combination is unusual at Nvidia's scale.

This is where investors need to be careful.

A 17-times forward P/E ratio sounds modest compared with the valuation multiples of some AI companies.

But valuation is only as good as the earnings forecast behind it.

If analysts expect 70% growth and the actual rate turns out to be 30%, the “cheap” multiple can suddenly look much less attractive.

That's why Nvidia's biggest risk may not be a collapse in AI demand.

It may be a slowdown relative to extraordinary expectations.

Investors don't need Nvidia to fail to make the stock disappoint.

They only need the future to be somewhat less spectacular than the market has already priced in.

Another reason the valuation debate is becoming more nuanced is competition.

Nvidia remains dominant in AI accelerators, but rivals are spending heavily to capture portions of the market.

AMD has gained significant ground.

Intel is rebuilding its AI ambitions.

Broadcom is supplying custom silicon to major AI customers.

And some of the world's largest technology companies are designing their own accelerators.

Reuters noted that Nvidia had been trailing several semiconductor peers on stock performance earlier in 2026 even while its fundamental business remained extremely strong.

That illustrates the difference between operating performance and stock performance.

A company can be doing extraordinarily well and still fail to outperform if investors believe competitors are improving faster.

There are actually several different debates happening under the same label.

One is whether AI technology itself is overhyped.

Another is whether AI companies are overvalued.

A third is whether AI infrastructure spending is being financed too aggressively.

And a fourth is whether Nvidia's stock price has detached from its fundamentals.

Those are not the same question.

Nvidia can be fairly valued even if some AI startups are not.

The AI industry can produce enormous economic value while individual investments turn out badly.

And hyperscalers can overspend on data centers without proving that Nvidia itself was a bubble.

This distinction is often lost in market commentary.

The strongest part of the bullish argument is not the P/E ratio.

It is cash.

The company is producing enormous amounts of operating cash as customers race to secure accelerators and related infrastructure.

That cash allows Nvidia to fund research, make strategic investments, build out its technology stack and repurchase shares.

The result is a business that increasingly resembles a mature technology platform in one respect and a hyper-growth company in another.

That combination is exactly why the stock does not fit neatly into the traditional bubble category.

Nvidia's valuation debate also reveals something larger about the AI boom.

The market is no longer driven only by companies promising future AI products.

Actual suppliers are producing enormous profits today.

Micron has just delivered record results.

Broadcom is seeing huge demand for AI infrastructure.

Data-center construction is accelerating.

Cloud companies continue making massive investments.

Nvidia is sitting at the center of that spending ecosystem.

That does not guarantee the boom will last indefinitely.

It does suggest the current AI cycle has a substantial real-economy component.

There are physical chips being manufactured.

There are actual data centers being built.

And there are customers paying billions for compute.

That is very different from an asset class whose value depends almost entirely on speculation.

The easiest mistake to make with Nvidia is staring at the numerator.

The company is worth trillions.

Its stock has risen dramatically.

Its AI dominance looks extraordinary.

But the denominator—earnings—is growing just as dramatically.

That is precisely why DBS's Hou argues the valuation does not resemble a conventional speculative bubble. Nvidia's forward multiple is much lower than those seen in many classic speculative episodes because the earnings forecast has expanded so rapidly.

The more important question is what happens when growth eventually normalizes.

No company can compound at 70% forever.

At some point, Nvidia becomes too large for that pace to continue mathematically.

The stock can still perform well after growth slows, but the valuation framework will change.

This is perhaps the most balanced conclusion.

Nvidia can be reasonably valued today.

The AI infrastructure boom can be real.

And the stock can still suffer a major correction tomorrow.

Markets move on expectations.

If investors suddenly conclude that AI capital spending will slow, the stock can fall even while Nvidia remains highly profitable.

Likewise, a change in interest rates can alter how investors value long-duration growth assets.

Geopolitical risks, export restrictions, supply constraints and customer concentration can also affect the outlook.

The absence of bubble-like valuation does not remove volatility.

It simply changes the reason investors should be cautious.

For now, the data support the argument that Nvidia is not obviously priced like the most extreme technology bubbles of the past.

A forward P/E around 17 times alongside projected earnings growth near 70% is difficult to compare with the speculative multiples of the dot-com era.

But the market has already recognized Nvidia's strength.

The challenge now is maintaining it.

The next earnings reports, new AI architectures, hyperscaler spending plans and customer demand will reveal whether the company's profit engine can continue expanding quickly enough to justify its enormous valuation.

That makes Nvidia's story less about whether AI is a bubble and more about something much more consequential:

Can the AI boom continue growing faster than Nvidia's already massive valuation?

So far, earnings have done an impressive job of catching up.

Wall Street is now betting they can keep doing it.