Investment implications of the AI bubble bursting

In yesterday's e-mail, I shared the many reasons why I have a bearish view on artificial-intelligence ("AI") spending, which I think is in a bubble, and especially on ChatGPT maker OpenAI.

Before sharing my investing advice related to this outlook, I'd like to share two thoughts...

1) The first comes from my old friend Alex Rubalcava, who explains why it's so difficult to predict which AI model(s) will come out on top, even as soon as a year from now:

First, regardless of who's got the hottest AI product this minute (now Claude Code, Gemini four months ago, etc.), all of them were trained on the previous generation of Nvidia Hopper chips. Nvidia only started to ship Blackwell chips a few months ago, and the models being built on the new chips are still in the middle of their first training runs. Blackwell chips have massively more compute than Hoppers do. So, true to "the bitter lesson," we can expect the performance of the models released in 2026 to be bigger leaps than the models of 2025, when there were no new chip families on which to train.

Second, most token consumption today goes to answering questions typed into generative systems, not to agentic systems. Generative AI competes for consumer time and attention with search engines, so there is a cap on the pricing that the foundation model companies can charge. Agentic AI, like Claude Code, doesn't compete with free alternatives for information retrieval. It competes with corporate salaries and budgets for getting work done, which means that they have pricing headroom that might be 10 times what's available in generative applications. They also use far more tokens than information retrieval, so the long-run unit economics are anyone's guess.

Alex concludes:

All that is to say that when pricing might be able to increase by 10 times, token costs are falling 10 times or more each year, and new models trained by the new generation of chips are just months away, it's nearly impossible to forecast unit economics for a company like OpenAI. Its business could be getting worse every month, or it could be getting better, and we just don't know. And whatever direction things are heading now, they could be heading in a different direction in three months!

Thank you, Alex!

2) My second thought relates to timing: I should have been clearer yesterday that I think the AI bubble could burst later in the year, not immediately.

Here's why: Most bubbles don't suddenly burst. If this one turns out to be anything like those we went through in 1999, 2007, or even late 2021, the biggest moves happen in the run's final phase, where there's a blow-off top before the crash. And that's where I think we are right now.

Sure enough, blowout earnings in the sector were reported overnight, as my friend Scott Tashman of Outset Global summarizes:

This morning, it's all about tech, as the sector benefits from a series of bullish headlines since the close. Earnings were mostly positive, with strong results and guidance from ASML, FFIV, SK Hynix, STX, and TXN.

ASML said the demand backdrop has improved meaningfully over the past two months, with blowout fourth-quarter bookings. STX delivered very strong results and said demand remains extremely strong, noting that it is already discussing calendar-year 2028 orders with customers. TXN's upbeat guidance pointed to improving conditions in non-AI end markets.

Beyond earnings, reports suggest two leading AI firms may pursue large fundraising rounds, with OpenAI seeking up to $100 billion and Anthropic potentially raising around $20 billion. Anthropic also reportedly increased its revenue guidance. Additionally, reports indicate China has formally approved H200 AI chip orders from some of its major tech companies, as expected.

Reader Jerome M. asks how we'll know when it's time to get out:

Do you have an estimate of OpenAI's runway to run out of cash, based on funds raised/burned, assuming they don't do a new round (SoftBank to the rescue...)?

It's a good question, especially in light of "OpenAI seeking up to $100 billion."

OpenAI obviously won't run out of cash as long as its backers shovel endless amounts into it. But when that stops, I think it will do so abruptly. (Note that "seeking" $100 billion is very different than "receiving" $100 billion.)

When (not if) OpenAI fails to raise the money it seeks, that will be a key indicator the bubble is bursting – and it will be time to get out.

3) So, if my forecast comes to pass, which stocks should you avoid and which should you hang onto or even buy?

The first stocks to collapse will likely be the smaller companies with insane valuations whose entire businesses (and, of course, stock prices) are dependent on AI.

One that comes to mind is CoreWeave (CRWV), which Google Gemini describes as follows:

It provides massive-scale GPU infrastructure designed for high-performance AI workloads, training, and inference. Partnering heavily with Nvidia, CoreWeave operates over 30 data centers, positioning itself as a key infrastructure partner for major AI labs and enterprises.

The stock has more than doubled since it went public last March. And as of yesterday, it has a $57 billion market cap, equal to more than 17 times trailing revenues.

It doesn't have a price-to-earnings multiple for this year because it's expected to lose money. But if you believe analysts' 2027 estimates of $2.26 per share (I don't), it's trading at nearly 50 times earnings two years out.

My friend and activist short seller Sahm Adrangi of Kerrisdale Capital published a bearish report on the company in September.

If you want to ride the AI bubble through the blow-off top, then stocks like CoreWeave will soar the highest – but also crash the fastest.

4) I think the implications of the AI bubble bursting will be very different for the Magnificent Seven companies. So I'd like to share my quick comments on each of them, in order of my most to least favorite...

Amazon (AMZN) and Alphabet (GOOGL) are tied for the No. 1 spot. I outlined the reasons why Amazon is my favorite in my November 5 e-mail, which still hold true:

The company has grown its revenues and profits at an exceptional rate over the past decade...

Yet, as you can see from this chart in my October 28 e-mail, its stock has done poorly over the past five years. Amazon has trailed its big-tech peers and the S&P 500 over that time frame...

[T]he stock is now trading at close to the lowest level of this [forward price-to-earnings] multiple in the past decade...

As I concluded, "I expect Amazon to follow (rapidly growing) earnings going forward." I also noted on January 7:

Amazon is also the No. 1 stock pick for 2026 of my friend and NYU marketing professor Scott Galloway – see: Big Tech Stock Pick of 2026: Amazon.

He argues that Amazon's massive investment in its warehouses – and especially robots – is about to pay off in the form of expanding margins and profits.

As for Alphabet, I wrote about its latest earnings report in my October 30 e-mail, reiterating that it's "a great stock for conservative, long-term-oriented investors."

And on January 12, I shared:

A Wall Street Journal article from last week highlights why I'm bullish on Google parent Alphabet and bearish on OpenAI.

As it notes, Google's Gemini AI app became the most downloaded app in Apple's app store this past September.

In November, Google launched its next Gemini model – the most powerful yet. And as the WSJ puts it, that meant Google "leapfrogged OpenAI to the front of the AI pack."

I think Alphabet is going to be the biggest AI winner – and one of my readers, Ramakrishna B., agrees:

My thoughts on Gemini: Google knows everything about me: all my browsing history, e-mails (Gmail), shopping history/purchases (via Gmail), search history, calendar, YouTube (videos and shorts), Google Home, Android OS, Google Maps, bookmarks in Chrome, Google Flights, etc. I use Google apps on my Samsung phone like Voice, Drive, Pay/wallet, Photos, etc. And I think this is true for many, many people!

This all provides massive amounts of data to teach Gemini, which is why I think it's become so good, so fast! Google has this unfair advantage over other AI tools.

I think this quote is correct: "It's over, bow down to your digital god (Google)."

No. 3 on my list is Meta Platforms (META), which I also think will be an AI winner. Its massive investments are paying off in better targeted ads.

I last analyzed the stock – along with Alphabet, Amazon, and Berkshire Hathaway (BRK-B) – in my December 2 e-mail, writing:

It boasts 3.54 billion daily active users of its various apps – including Facebook, Instagram, and WhatsApp (none of which I think will be displaced).

Despite its size, last quarter it grew revenues by a remarkable 26% year over year. (I also covered Meta's latest earnings in my October 30 e-mail.)

Next, I'm neutral on Apple (AAPL) and Microsoft (MSFT). They're both insanely great, cash-generating companies, but I'm cautious on them for different reasons regarding the AI bubble bursting.

On June 18, I wrote that "Apple is priced as if it's a growth company... but it's not." My view hasn't changed.

As for Microsoft, it has invested roughly $13 billion in OpenAI and owns around 30% of the company. It's not a huge part of Microsoft's $3.6 trillion market cap, but if OpenAI blows up, it could hit Microsoft's stock.

As for Nvidia (NVDA), this company is at the epicenter of the AI boom, so its stock could easily be cut in half if the bubble bursts. If you own it and have big gains, consider holding to possibly catch a last upward move but also having a tighter stop loss to avoid giving all those gains back.

Lastly, my least favorite stock in the Magnificent Seven by far is Tesla (TSLA) – but for reasons unrelated to a possible bursting AI bubble.

I even included it in my "Stinky Six" list of stocks to avoid, which I first published on October 29. (I'm pleased to report that every stock on the list is down since then – including Tesla, down 7% – with an overall average return of negative 15%. The S&P 500 Index is up 2% in the same time frame.)

As I wrote then:

With its tax-credit expiration in the U.S. and competition from Chinese manufacturers, I expect Tesla's vehicle sales to collapse. I'm also very skeptical that the company's Optimus humanoid robots and the nascent Robotaxi business will take off.

In addition, its valuation is absurd: With a $1.4 trillion market cap, the stock trades at 265 times 2025 earnings per share and 198 times 2026 estimates (which I don't believe).

So, in summary...

Amazon, Alphabet, and Meta remain my three favorite large-cap tech stocks.

I'm more cautious about Apple and Microsoft. But that doesn't mean dumping the stocks if you own them.

Nvidia doesn't look like an outright "sell"... But it could be a good idea to lock in some profits if you have the opportunity.

And my view on Tesla remains the same as it has been – stay away from the stock.

Best regards,

Whitney

P.S. I welcome your feedback – send me an e-mail by clicking here.