When Founders Come Back, Pay Attention
Here's a pattern worth knowing. Steve Jobs came back to Apple in 1997 — the stock was a penny joke. Howard Schultz came back to Starbucks in 2008. Michael Dell took his own company private in 2013 when everyone else had written him off.
Every time, the same thing happened. The founder shows up, the company changes, and the people who were paying attention made money.
So it's worth noticing that Jeff Bezos — still Amazon's Executive Chair — has quietly re-engaged with what the company calls "large-scale" initiatives. Meanwhile CEO Andy Jassy is lighting a bonfire of cash on the AI buildout: roughly $200 billion in capital spending planned for 2026, most of it going into AWS and AI infrastructure. AI revenue inside AWS is already running at a $15 billion annual pace. Amazon's own custom chip business? $20 billion run rate, growing triple digits.
You don't drop $200 billion on a hunch. And founders don't quietly re-engage unless they think the next chapter is bigger than the last one.
The $26 Trillion Question — How Big Is This, Really?
Whenever somebody throws out a number like "$26 trillion," your first instinct should be... skepticism. Fair enough. So let's look at what the grown-ups are actually forecasting.
The honest answer: nobody knows. But here's the range from the most-cited research shops.
Notice how the numbers spread out. The "market" forecasts cluster around $2–5 trillion. The bigger "$15 trillion plus" figures include downstream productivity gains across the whole economy — every industry, every job. That's where the $26 trillion headline comes from. It isn't made up, but it's also not a stock-market number. It's a civilization-scale guess.
Meanwhile, the money being spent right now is very real. Goldman Sachs puts 2026 hyperscaler capex at $527 billion, up from $394 billion last year. That's more than the entire defense budget of most countries.
Is This 1999 All Over Again?
Look — if you lived through the dot-com crash, you've earned the right to be skeptical. So here's where we are.
The S&P 500's long-term P/E ratio is at its highest level since 1999.
In November alone, news stories mentioning "AI bubble" equaled the prior ten months combined.
Bank of America's December fund-manager survey called the AI bubble the biggest tail risk in markets.
Michael Burry (yes, that guy from The Big Short) and the Bank of England have both sounded the alarm.
And there's this weirder thing: circular money. Nvidia invests in Oracle. Oracle commits to buy Nvidia chips. OpenAI signs a $300 billion deal with Oracle. Meta, Alphabet, and Oracle together need to raise around $86 billion in the bond market this year just to keep up. It rhymes with stuff we've seen before, and it isn't flattering.
But — and this matters — it's not 1999. The companies leading this wave make actual money. Nvidia's profits are real. Meta's are real. Microsoft's are real. These aren't pets.com with a ticker. Debt-to-earnings levels are lower than the WorldCom era. Bezos himself called this an "industrial bubble" — meaning yes, a lot of capital's gonna get torched, but the infrastructure left behind will transform the economy.
Translation: the hype is overheated. The technology isn't.
Where the Next Gains Are Actually Hiding
Here's the thing most people miss. The Magnificent Seven — Apple, Microsoft, Alphabet, Amazon, Nvidia, Meta, Tesla — now make up roughly a third of the entire S&P 500. If you own an index fund, you already own a giant pile of them. You don't need more.
The smart money has been quietly rotating into the companies that make the AI boom physically possible — the "picks and shovels" crowd. And the action has moved past GPUs.
These are the companies supplying the memory, the networking gear, the cooling systems, and even the electrical contracting to wire up data centers. Unglamorous. Essential. And largely "silicon-agnostic" — meaning it doesn't matter which AI model wins. They all need the plumbing.
The Playbook from Here
So what do you actually do with this?
Check your real exposure. If you own an S&P 500 index fund, roughly 30% of it already sits in seven AI-heavy stocks. That's concentration, not diversification.
Don't pay stupid prices. Anything above 30x forward earnings assumes the AI money printer runs on schedule. It might. It might not.
Keep 2–3 years of expenses safe. Short-term Treasuries, money market funds, CDs. Whatever helps you sleep. Yale and independent analysts are modeling potential drawdowns of 20–30% in AI-heavy names over 2026–2027.
Balance with the "real economy." Healthcare, energy, quality dividend payers, select mid-caps — stuff whose earnings don't depend on the next Nvidia earnings call.
Remember the Amazon lesson itself. During the dot-com crash, Amazon stock fell more than 90%. The company survived, thrived, and became one of the most valuable businesses in history. But the investors who got wiped out in the fall didn't stick around to enjoy the ride back up.
The AI revolution? Real. The AI bubble, in parts? Also real. Your job isn't to pick between them — it's to be positioned for both.
If you're done with hype and want straight talk on markets, politics, AI, and tech - subscribe to Trade Talks Live.
