Years ago, my younger brother played football for Lehigh University. My athletic career was over by then, so I had a great time living vicariously through him.

As a senior, he was a team captain and All-Patriot League. I still remember going to his games. And I have fond memories of the larger Bethlehem, Pennsylvania, area the school is situated in.

For one thing, the Christmas decorations there are top notch. For another, the town hasn’t lost sight of its unique steel mill heritage, a legacy I respect.

Though its claim to fame, Bethlehem Steel, was dissolved in 2003 after filing for bankruptcy, it’s still quite the story to tell. Founded in its namesake town back in 1857, the company went on to become the second-largest steel manufacturer in the United States.

Today, its old factory is a casino-resort filled with neon lights and thousands of gaming tables. But outside, you can still see a lot of the old mill components.

Source: Art Quest

For instance, nearby is the Steel Stacks campus, where art shows, music festivals, and other community events are held throughout the year. It’s interesting to tour the grounds and look at all the old contraptions that helped build America.

Plus, they light up the old blast furnace towers at nighttime. It’s pretty cool to see the past come alive in a more hip, modern setting.

Which brings me solidly from the 20th century to the 21st. Because when I think about Bethlehem Steel today, I oddly find myself thinking about Nvidia (NVDA).

Yes, that Nvidia. The Big Tech company that just reported record Q2 results – including:

  • $96.2 billion in sales, up 105.9% year over year

  • Net income of $59.7 billion, up 126%

  • $26 billion returned to shareholders via stock buybacks and dividends.

There are a few different reasons why…

Nvidia’s fundamental worth isn’t showing in its shares

Back in Bethlehem Steel’s day, the world was running on the product it and its industry made. There’s no way to understate how important that resource was in building the reality we take for granted.

Steel created hundreds of thousands of jobs both directly and indirectly, and facilitated countless others. Skyscrapers wouldn’t exist otherwise. Nor would large bridges, railways, or highways.

Heck, we probably wouldn’t have won World War II without the steel produced in Bethlehem.

This is just like how artificial intelligence (AI) is transforming our world today.

Nvidia is actually the biggest beneficiary of the global data center buildout. Data centers are the new-age steel mills. And Nvidia is powering them with its GPUs (graphics processing units).

During its most recent quarter, the company produced $89 billion in data center revenue.

That was up 117% year over year!

All told, Nvidia just posted financial results in a single quarter that outclass what the vast majority of S&P 500 companies can produce in an entire year. Yet it’s still getting too little market love, with its earnings multiple compressing by the day.

Looking at forward earnings expectations, Nvidia is trading for less than 18x earnings as I write this… which is below the S&P 500’s forward price-to-earnings (P/E) ratio of approximately 20x.

Source: Chat GPT

It’s just so odd to me to see a company posting triple-digit top-line growth with 75% margins… yet trading at a sub-market multiple. This doesn’t make fundamental sense.

Investors fear the worst for Nvidia

There’s no doubt in my mind that Nvidia deserves a premium valuation. However, investors aren’t willing to pay it for one simple reason.

They’re afraid of a data center slowdown.

Nvidia’s management team did its best to quell these concerns during the company’s recent quarterly report. It highlighted Q3 guidance, which calls for $108 billion in sales with a 74% gross margin rate.

In which case, it’s clear that Nvidia’s growth story won’t end in 2026… even if that’s precisely how it’s being valued on the market.

In those naysayers’ defense, no company is impervious to failure. Going back to Bethlehem Steel, it wielded power for well over a century yet still ended up going bankrupt.

Then again, that’s just as easily a point in my argument’s favor considering how that company rode a wave of global, game-changing demand for so long. Nvidia is only 33 years old and three or four years into the tsunami of AI demand.

The way I see it, it still has a long way to go.

Yes, I also recognize that many people think Nvidia is playing a dangerous game of musical chairs. They think that when the data center buildout music stops, the company’s fundamentals fall apart.

First off, let’s return to the fact that Nvidia shares are trading at less than 20x earnings. If they were trading for 50x earnings, then sure. I’d agree the market’s entire focus should be on future data center growth.

That sort of valuation would require massive forward-looking growth to justify it.

However, as it is, the company doesn’t need to experience groundbreaking growth to generate solid returns for its shareholders. In which case, investors shouldn’t be asking how many more data centers will be built. They should be asking how Nvidia will monetize the existing market going forward.

This is where the steel stacks matter once again…

The Nvidia naysayers aren’t seeing the whole picture

Historically, when someone built a steel mill, they knew that the heavy parts they were assembling would last for decades – if not hundreds of years. That’s why the Bethlehem landscape along the Lehigh River looks the way it does today.

The same thing can be said of other big infrastructure developments. Take oil refineries, where the last major one built in America was completed in 1977.

Yet the U.S. still refines roughly 17 million barrels of oil per day, accounting for a fifth of the world’s crude oil supply.

Sure, there’s maintenance needed. Some parts break. Others need to be updated.

But, in general, once the initial infrastructure was built, their owners could rely on it year in and year out. Data centers, however, won’t work like this.

The shell of the buildings, at least, should last for decades. But because of the amazing rate of innovation we’re seeing in the semiconductor industry, it’s unlikely that the chips – which currently account for about 50% of a data center’s cost when being built – will be relevant for about five or so years.

That means the hyperscalers and other data center owners will need to go back to Nvidia at least once a decade with major orders to refresh their existing portfolios.

I’m sorry, but I don’t see these companies – which have already invested hundreds of billions of dollars on data centers – letting their assets become obsolete. Even if they have to cut back on building new facilities, their need to maintain existing ones should result in huge, reliable, and predictable sales for Nvidia…

Turning it into a toll booth of sorts when it comes to compute.

There are already more than 4,000 data centers operating in the U.S. right now. Globally, this figure doubles; and by 2030, there’s little doubt it will climb above 10,000.

That in and of itself is a huge market for chip re-fresh sales. Any growth past it is just a cherry on top of Nvidia’s durable-demand ice cream sundae.

This is why this tech giant had the confidence to raise its dividend by 2,400% back in May…

Why I believe it will become one of the most generous companies in terms of shareholder returns over the coming decades…

And why I see its shares as being extraordinarily cheap right now.

Why the world’s best company is trading at a sub-market multiple is beyond me. Yet that’s the reality we live in.

All I can say is that, as a long-term investor, I’m happy to take advantage of this discount while it lasts.

Kind regards,

Nick Ward
Analyst, Wide Moat Research

Disclosure: Nicholas Ward is long Nvidia (NVDA).