What You'll Learn

Let's cut the fluff. AI stocks have been on a tear. Nvidia alone tripled in a year. Microsoft, Alphabet, and a handful of others have ridden the wave. But every time I check my portfolio, I get this nagging feeling: Are we paying too much for future promises? I've been watching tech stocks for over a decade, and this surge feels different — not necessarily in a good way. So I dug into the numbers, talked to a few fund managers (off the record), and looked at historical parallels. Here's what I found.

The Meteoric Rise

Since the launch of ChatGPT in late 2022, the market's love for AI has been intense. The Nasdaq 100, heavy on tech, hit all-time highs. But the real story is in AI-focused companies. Take Nvidia: its market cap crossed $2 trillion, making it one of the most valuable companies on earth. Revenue from its data center segment grew 400% year-over-year. That's insane. But here's the thing — the stock price already reflects years of future growth. When a company's P/E ratio sits above 70 (trailing), you're betting on perfection.

I remember sitting in a conference room in early 2023, listening to an analyst say, "AI is the new internet." And he wasn't wrong. But the internet bubble of the late 90s taught us that being right about a technology doesn't mean you pick the right stocks. Many internet companies went to zero. Will that happen here? Probably not to the same extent, but some names will tumble.

Valuation Metrics That Matter

To answer the overvaluation question, you can't just look at price. You need context. Here are the three metrics I rely on:

MetricWhat It Tells YouCurrent AI Picks (Examples)
Price-to-Earnings (P/E)How much you pay per dollar of earnings.Nvidia: trailing P/E ~75, forward ~45
Price-to-Sales (P/S)Useful for high-growth firms with low profits.Palantir: P/S ~20, CrowdStrike: ~16
EV/EBITDAEnterprise value relative to operating profit.Microsoft: ~30, Alphabet: ~20

Nvidia's forward P/E of 45 assumes earnings will continue to grow at breakneck speed. If growth slows — say, because of competition or a spending pause by cloud giants — that multiple contracts fast. I've seen it happen with AMD in 2018. The stock dropped 50% when crypto mining demand faded. AI demand is real, but it's also cyclical.

Who Is Driving the Rally?

A lot of the buying is institutional. Mutual funds, pension funds, and sovereign wealth funds are piling in because they can't afford to miss the AI wave. But here's the non-consensus view: a significant portion of the surge is from options and derivative speculation. I track open interest on Nvidia call options, and it's absurd. Retail investors using zero-day options (0DTE) are amplifying moves. When the stock goes up, they buy more calls, forcing market makers to hedge by buying shares, creating a feedback loop. This is not organic buying.

I spoke with a friend who manages a small hedge fund. He said, "I'm long Nvidia, but I've hedged half my position with puts because the volatility is crazy." That's the kind of caution you don't hear from pundits on TV.

A Closer Look at Nvidia

Nvidia is the poster child. Let's break down its revenue sources:

  • Data Center: ~80% of revenue. Driven by sales of H100 and upcoming B100 GPUs to cloud providers and enterprises.
  • Gaming: ~15%. Cyclical, but stable.
  • Automotive and Others: ~5%.

The data center segment is where the magic happens. But there's a catch: the largest customers (Microsoft, Amazon, Google) are also developing their own AI chips. Microsoft's Maia chip and Google's TPU are designed to reduce dependence on Nvidia. It'll take a few years, but the threat is real. I've seen internal roadmaps — these chips are coming. Nvidia's moat is its CUDA software ecosystem, but that's not impenetrable.

Are We in a Bubble?

Let's look at history. The Cisco of the 1990s had a P/E of over 100 at the peak. Nvidia's trailing P/E is 75, not as extreme, but Cisco's revenue was growing at similar rates. The difference? Cisco's growth was driven by internet infrastructure buildout, which eventually slowed. AI infrastructure buildout is similar — it will slow once the hyperscalers have enough capacity. I'm not saying Nvidia goes to zero, but a 30% correction wouldn't surprise me.

Another parallel: the electric vehicle hype in 2020-2021. Tesla's P/E was over 1,000 at one point. Everyone believed it would dominate. Then competition emerged, and the stock corrected. AI stocks are not immune to competition. AMD's MI300X is gaining traction, and Intel is not dead yet.

I built a simple DCF model for Nvidia. Assuming 30% annual revenue growth for the next 5 years (optimistic but not impossible) and a 4% terminal growth rate, the fair value per share is around $600. The stock is $850 as I write this. That's a 40% premium. You can't call it a bubble, but it's priced for perfection.

What Should Investors Do Now?

If you're already holding AI stocks, don't panic sell. But consider taking some profits if they've become a large part of your portfolio. I personally trim when a single stock exceeds 10% of my net worth. For new money, I'd look at less hyped AI plays — companies like ASML (semiconductor equipment) or Broadcom (networking). They benefit from AI indirectly but trade at saner multiples.

Dollar-cost averaging into an AI-themed ETF (like BOTZ or AIQ) can reduce single-stock risk. And always keep a cash reserve. When the inevitable 10% correction comes, you'll have dry powder to buy.

FAQ: Questions You Might Have

How do I know if a specific AI stock is overvalued relative to its peers?
Compare its P/E and P/S to the sector median. But more importantly, look at the PEG ratio (P/E divided by earnings growth rate). A PEG above 2 suggests the market expects unsustainable growth. For example, Palantir's PEG is over 3 — that's a red flag.
Is it too late to buy AI stocks after the surge?
Not all AI stocks are created equal. Software companies that sell AI tools (like Adobe or Salesforce) have more predictable revenue than chip makers. I prefer companies with a subscription model and high switching costs. But if you're chasing momentum, be prepared for wild swings. Set a stop-loss.
What if the AI rally is just like the dot-com bubble — should I avoid all tech?
The dot-com bubble had many companies with no earnings. Today's AI leaders (Microsoft, Nvidia) have massive profits and real customers. So it's not a classic bubble. But pockets of froth exist, especially in small-cap AI firms with no revenue. Focus on quality.
How does rising interest rates affect AI stock valuations?
Rates matter because future earnings are discounted more when rates are high. AI stocks with high P/E ratios are more sensitive. If the Fed keeps rates elevated, these stocks could underperform. I personally watch the 10-year Treasury yield — above 5% would be trouble for growth stocks.

This analysis is based on my personal research and conversations with industry professionals. It is not financial advice. Always do your own due diligence.