While many investors are earning huge returns on stocks connected to artificial intelligence technologies, a growing chorus warns that this boom has strong echoes of the dot-com bubble of the 1990s.
For many investors today, the internet bubble is only something they’ve heard or read about. But Dan Chung lived it as senior tech analyst at Alger Funds, charged with following hot New Economy stocks like Yahoo. Then the dot-com bubble popped. Shortly afterward, Chung became Alger’s chief investment officer as the firm was decimated by the World Trade Center attack. Alger rebuilt with its signature strategy of identifying disruptive growth companies. Today, Chung is CEO of Alger, which oversees $33 billion. He co-manages an array of funds, including Alger 35 ETF ATFV, which is up 51.83% over the past 12 months, versus 22.87% for the Morningstar US Large-Mid Broad Growth Index.
In a recent conversation, Chung told us why he thinks this market, and the AI boom, have plenty more room to run. He explains why he thinks the companies driving this rally are more solid than those in the dot-com era, and he shared a few of his favorite stocks. This conversation has been edited and condensed for time and clarity.
Leslie Norton: People are comparing the AI-driven stock market with the dot-com bubble. What do you think?
Dan Chung: Somebody once asked the late David Alger, “What do you do when you see a bubble forming?” And I think David famously said, “Well, first I jump in.” A bubble would imply we’re irrational in some way. That’s not correct. I don’t think we’ve gotten there yet.
I look at four big things. One, what the market is saying in terms of market behavior, momentum, and the returns of stocks associated with the thematic trend, which is AI. Two, the fundamentals of the leading companies, dissociated from valuation. Three, valuation, which we look at through scenario-based, upside return potential to downside risk. Four, the macro environment for the industry and the broader economy.
The fastest growers—the stocks associated with AI—have had a ton of momentum in the last nine months. It hasn’t reached 1998-99 levels yet. And you’re talking to one of the few who was a senior analyst leading tech in the 1990s. I would say we probably still have at least 1998 and 1999 to go through, which is 50% more than today. We didn’t do so great on many internet stocks back then, but I remember selling Yahoo at the very top in December 1999. I also remember how that felt, what the momentum and valuations looked like. And we’re not there yet.
Norton: Where are we, then?
Chung: We’re in the middle stages of the boom. You don’t want to miss the second half. Missing the second half of the dot-com boom killed [celebrity investor] Julian Robertson and the original Tiger Fund.
Microsoft Then and Now
Norton: Let’s revisit some other comparisons.
Chung: So many companies no longer exist, or are unrecognizable. One easy way to compare is using Microsoft MSFT. Microsoft hasn’t really changed that much in terms of relative position and products, but it’s the biggest software company in the world. At the very top of the dot-com bubble, Microsoft traded at 67-70 times earnings and at a P/E ratio three times the market. Today it’s only trading at 1.5 times the S&P 500, which is about average for a full 30-year period. Its P/E ratio is 32 on consensus estimates. While 32 is elevated, it’s nowhere near 67 or 70 times.
Norton: How about fundamentals?
Chung: Even back then, Microsoft was a profitable, leading company. It is even moreso today. Revenue growth, margins, and free cash flow are vastly superior to what they were at the top of the dot-com bubble. We analyzed the capex of the hyperscalers—Microsoft, Google GOOG, Meta Platforms META, Amazon AMZN, Oracle ORCL. These five collectively have $1.1 trillion in revenue and around $550 billion of operating cash flow in 2025. Their capex spending is about $400 billion, and roughly half of that is AI spending. So total free cash flow after all capex is about $160 billion, or $360 billion excluding AI capex spending. All those numbers are way better than the leading stocks of 1999. Yahoo was trading at 350 times earnings and the stock was going ballistic.
The economy was fine in 1998-99. It looks fine today. Interest rates were higher back then. From 1998, rates were forecast to go up, and the average 30-year fixed-rate mortgage in 1999 was 7.5%. Right now, the bond market says rates will go down. We’ll see. That would obviously be supportive for equities as well as the economy. Stocks were even more overvalued in 1999, because at those interest rates, they should have had lower P/Es.
Norton: At the time, burn rate was a big concern.
Chung: Amazon then was completely unprofitable. People who focused on the short term and whether Amazon was cash flow positive did not understand the business model and missed the stock. To be arguing about cash flow margin, and weighing that as more important than top-line growth and potential for growth, was clearly a mistake. My point is, the financial metrics of the leaders back then look terrible compared to today’s AI leaders.
Maybe it’s not a good sign about our economy that the AI leaders are some of the companies that were biggest and best over the last 15 years. That wasn’t true in in 1999, when the biggest stocks in the S&P 500 included names like Costco COST, GE, and Exxon XOM. None were internet stocks. The only real internet stock was Microsoft. And the internet leaders included Amazon, eBay EBAY, Yahoo, and Cisco Systems CSCO.
When we talk about the AI leaders, it’s Nvidia NVDA, Microsoft, Amazon, and Google, all the hyperscalers. Some of the other leading AI companies are private, like OpenAI. Tesla TSLA is a clear AI player in autonomous driving, its humanoid robot program, and its AI effort. Apple AAPL hasn’t executed, but it obviously has opportunity. They’ve been big, successful companies for the past decade, and that wasn’t true with the internet bubble.
Nvidia vs. Cisco
Chung: By the way, Cisco’s P/E at the top was 126 times earnings. In mid-1999, its P/E was actually very similar to Microsoft’s, in the 60-70 range.
Norton: How does that compare with Nvidia today?
Chung: That’s just it. Nvidia’s P/E right now is 33.5 times consensus earnings. Our estimates are considerably higher, so obviously, for us, it looks cheaper. People keep saying Nvidia has gone up so much and cite Newton’s law. But it’s all tied to the fundamentals, and Nvidia’s fundamentals are extremely strong. This year, it’s going to grow nearly 50% on the top line. They’re going to grow earnings by 65% and free cash flow by 67%. And the price-to-earnings growth ratio is actually under 1.
Norton: What concerns you?
Chung: The good thing is these companies have a trillion dollars of revenue, half a trillion of operating cash flow. However, the hyperscalers have increased the capital spending 60%-80%, building out data centers, buying Nvidia chips, upgrading electricity. They’re still free cash flow positive, which just shows you how rich these companies are. Will all this capital earn a reasonable return? Time-wise, we’re very early. ChatGPT burst onto the world only three years ago in January. AI can touch everything. You don’t really know if they will make ROI on these capital investments. How far will this permeate our working and personal lives?
We’ve done a lot of work triangulating with different methodologies (some that are almost too nerdy) to assess the potential size and ROI of the AI market. Is the total addressable market big enough? One way to think about it is to model what jobs are susceptible to AI replacement. For software programming, 60% could be affected. For road construction, 5%. And then, if AI saves a business $100 of cost, you’re willing to spend $50 for that. That’s a two-year return on investment, so that may actually be conservative.
So is [Nvidia CEO] Jensen Huang nuts when he says the market for AI tech, services, hardware, computing could be $2 trillion-$3 trillion in 2030-31? No. We can get to his numbers on reasonable assumptions about AI replacing workers through enhanced productivity without, for example, assuming 100% of people become unemployed because of AI. The bulls like Jensen are aggressive but not outlandish.
Why Chung Sold Yahoo During the Dot-Com Bubble
Chung: When we sold Yahoo, I loved the revenue growth. It was the dominant search engine. But even on my most bullish fantasizing, I could not justify its 350 P/E ratio. I wondered if I should use a higher multiple or higher margins. But I couldn’t get there. That was a lucky sale in a lot of ways, but it was informed by that kind of thinking.
Norton: Let’s talk more about your outlook.
Chung: Everybody’s looking at the tariff impacts and the trade war. I believe we’ll see the tariffs filter through to higher prices in January, February, March. I think we’re starting to get to the end of the inventories. We see European merchants raising prices 10%, citing tariffs. In our proprietary research with industrial companies, they’ve said their actual cost is only going up 4%-5%. Some are eating that to gain market share. Some are aggressively raising prices because they think the clients have no choice. I don’t think we’ll have skyrocketing inflation.
Here’s the negative: While the economy looks like it’s doing quite well, it’s companies exposed to upper-end consumers. It will be interesting to see if the Trump administration’s reshoring efforts make a difference and create a much better job market for Americans and legally resident foreigners. A lot of people question whether the American workers will be as cost-efficient as Asian or Chinese workers. That price will have to be paid by all of us, but I’m open-minded. Potentially, it may lead to better, overall, more balanced economy and society. But it’s creating a new deck of billionaires in a business poised to unemploy a lot of people.
Chung’s Stock Picks
Norton: What stocks do you like?
Chung: Every year around this time, everyone wants to know if it’s time to sell Nvidia because it was up so much. It’s not the time to sell Nvidia. I gave you the multiples and the growth rate. There’s still a lot of upside. The stock chart has basically done nothing since July. A lot of people have gone into more meme-like stocks that have skyrocketed and ignored this high-quality company with great earnings, the absolute leader in this technology.
Norton: What else?
Chung: Nebius NBIS is the leading AI data center company. Our team, led by Ankur Crawford and Patrick Kelly, identified it really early, and the stock has broken out. In the last year, it has announced major deals with the hyperscalers. We think it will grow tremendously over the next 10 years. The company has multiple technology and other investments, all of which will benefit from AI.
They have a stake in an open-source data management business called ClickHouse, used by software programmers and developers. It’s a very important, high-quality development platform. It has an autonomous driving technology developed years ago, when Nebius was part of Yandex. It left Russia, reorganized and reformed in Europe, negotiated the loss of the Russian business, but retained other assets, including its autonomous driving technology.
It’s a bit like Softbank, in that they have a core business, like data centers and other businesses that they invested in, grew, and developed. We don’t think the market fully values it. Nebius is growing 400% this year and another 400% next year as well. We’re also looking at it becoming EBITDA-positive next year. We do see it as an extremely high grower five years out. We see Nebius going from over $500 million in revenues in 2025 to over $2.5 billion in 2026. The stock basically should double over the next three to four years.
