sport football: Will the AI Bubble Pop in 2026?

July 22, 2026

Teeka Tiwari

Friends,

It’s the question on everyone’s mind…

“Is the AI bubble about to pop? Is a major crash imminent?”

Get my answer here…

Because it’s critical for you to act before August 14.

Let The Game Come To You!

Big T

In case you missed it, here’s Big T’s Digital Asset Daily

On June 10, two days before SpaceX went public, I made a simple call:

SpaceX is extraordinary… I’m not betting against this company or [Elon Musk]. What I am betting on is this: you’ll be able to own shares at a MUCH cheaper valuation in the future.

I didn’t need a crystal ball to tell you that. I just needed to show you the math. (More on that below.)

SpaceX shares hit a peak of $225.64 a few days after its IPO. As of Tuesday’s close, they were trading at $123.54. That’s a 45% drop from the peak.

SpaceX priced its IPO at $135 a share. Tuesday’s closing price sits 8% below that. So even the buyers who got in at the IPO price are sitting on a loss. Not just the ones who chased it at the top.

I’m not saying SpaceX is a bad company. I’m not saying Elon Musk won’t go on to build one of the most dominant businesses in the world.

I’m saying price matters. And during the IPO frenzy, SpaceX never gave you a good one.

So why did this happen? I ran the numbers for you back in June.

SpaceX was targeting a $1.75 trillion valuation. That’s almost 94x its revenue, for a company that lost nearly $5 billion last year. Only about 4-5% of its shares were available to trade at listing.

When you combine sky-high hype with a tiny float, you get a stampede of buyers chasing very few available shares. The price gets bid up past any reasonable level.

The early excitement fades, the float eventually expands, and latecomers get stuck holding the bag at prices that no longer make sense.

Don’t get me wrong. Elon Musk built something remarkable with SpaceX. But even a remarkable company can lose you money if you pay too much for its shares.

The Trade That Actually Matters

Avoiding SpaceX at the wrong price doesn’t mean ignoring the bigger trend behind it. That’s the mistake I don’t want you to make.

Because the real story was never just SpaceX’s rockets. It was what SpaceX became once AI entered the picture.

After merging with Elon Musk’s xAI, SpaceX is no longer just a space exploration company. It is now tied directly to the AI data center and infrastructure buildout.

And that buildout is massive.

McKinsey estimates global spending on data centers could reach nearly $7 trillion by 2030. Morgan Stanley forecasts U.S. data center power demand could hit 74 gigawatts by 2028, with a shortfall of about 49 gigawatts in available power.

That gap is roughly equivalent to the entire average power consumption of Florida, the third-largest electricity-consuming state in the country.

And every one of those data centers needs the same thing: Power.

It doesn’t matter whether xAI, Meta, Google, Amazon, or Microsoft wins the AI compute war. Every one of them needs electricity to train models, run servers, and keep data centers operating around the clock.

So the thesis behind the AI power trade hasn’t changed. If anything, I believe it’s gotten stronger.

Big Tech Is Starting to Look Like Big Oil

Last week, we got proof of just how brutal the fight over the AI model layer is becoming. Meta fired the first real shot in what I’ve been calling the “AI spending war.”

The company priced access to its new AI model, Muse Spark 1.1, at $1.25 per million input tokens and $4.25 per million output tokens.

Tokens are how AI companies measure usage. A token is usually a piece of a word, a full word, or a small phrase. Input tokens are what you feed into the model — your questions, prompts, commands, or instructions. Output tokens are what the model gives back — the answer, analysis, code, image instructions, or response.

Now, compare Meta’s price to Anthropic, which charges $5 and $25 per million tokens for its top AI model. That’s 4-6x cheaper.

When the biggest players start slashing prices like that, it means the AI compute business is becoming a commodity, just like coal or natural gas.

And that’s the problem. The industry is spending more to produce the product while charging less to sell it.

That’s a recipe for lower stock prices.

This isn’t just a Meta story. A recent Bank of America report showed the same pressure building across the entire hyperscaler group. The five biggest hyperscalers have spent $234 billion on AI infrastructure so far this year.

Their stocks, as a group, are basically flat in 2026. Their free cash flow, the money left over after paying the bills, is falling. Meanwhile, the companies supplying the equipment are seeing their cash flow climb.

Bank of America called it “a generational transfer” in free cash flow.

I’ve been saying this for months, and I’ll say it again: The AI hyperscalers may still produce winners, but that layer is getting more competitive, more expensive, and more price-sensitive by the month.

Open-source models are catching up fast. Once compute costs come down further, there’s no good reason to keep paying premium prices for a closed model when a free one does the job almost as well.

Look, Big Tech hasn’t become Big Oil. But part of its business line is starting to look a lot more like it: expensive to build, brutally competitive, and increasingly hard to operate at high margins.

Think about what that does to the valuations.

Exxon is one of the biggest energy companies on the planet, and it trades at roughly 2x sales. Meta, Microsoft, Alphabet, and Amazon trade at an average of 7x sales. They’re priced like growth companies, not commodity companies.

I’m not saying these companies are destined to trade at 2x sales. They’re different businesses from energy.

But if investors start valuing even part of their AI spending like a commodity business – with lower margins, heavier capital costs, and tougher competition – their multiples could suddenly compress.

The historical baseline for a mature, commoditized tech giant is a sales multiple of 4x to 5x. That matters because Wall Street is not pricing these companies as mature, commoditized businesses today.

Combined, the four major hyperscalers trade at a market cap of $11 trillion. If their average sales multiple compresses by just one-third – from 7x to 4.7x – you’d see nearly $4 trillion in value disappear just like that.

Profit From the Other Side of the Power Bill

Meta can slash the price of its AI computing all day long. It cannot slash the price of the electricity running its servers. That bill gets paid, no matter who wins the AI compute war.

That’s why we’re looking for companies on the other side of that power bill.

Recently, my team uncovered a power producer with long-term contracts already signed to sell electricity directly to the biggest names in AI.

It generates steady, around-the-clock power, the exact kind a data center cannot run without. And it trades at a fraction of the valuation Wall Street is paying for the popular AI stocks everyone’s chasing.

Based on our research, we believe this stock could see as much as 261% upside from here.

I published the full write-up, including the name, the ticker, and the contracts behind it, in the July issue of my flagship research service, Asymmetric Edge. If you’re already a subscriber, you can read it here.

If you’re not a member yet, I recorded a briefing that walks through this thesis and the companies positioned to profit from it.

Friends, SpaceX provided the lesson I wanted you to see before the IPO: even a great company can be a bad investment at the wrong price.

The real money is made by owning the right companies at the right price — especially the ones that get paid no matter who wins the AI war.

Let the Game Come to You!

Big T