In Berkshire Hathaway’s 2018 shareholder letter, released in February 2019, Warren Buffett showed investors exactly how powerful buybacks can be. |
Eight years earlier, Berkshire owned 12.6% of American Express. Berkshire didn’t buy a single new share in that time. Yet its ownership stake had grown to 17.9%. |
How? Buybacks. |
Every time American Express repurchased its own stock, the number of shares left in the market shrank. Berkshire’s slice of the company grew bigger without spending another dime. |
Buffett put it simply: “When earnings increase and shares outstanding decrease, owners, over time, usually do well.” |
And the pattern didn’t stop there. |
By the end of 2025, Berkshire still owned the same 151.6 million AXP shares. But because American Express kept shrinking its own share float, Berkshire’s percentage ownership of the company had climbed again, to 22.1%. |
That’s the power of a shrinking share count. |
Every buyback grows your proportional ownership and your claim on the company’s earnings, for as long as management keeps repurchasing shares without overpaying for them. Buffett himself only liked the trade when he believed the stock was underpriced. |
That’s why buybacks matter so much to your bottom line. And it’s exactly why what I’m about to show you about AI hyperscalers should worry you. |
The Deal That Built Your Portfolio |
For years, the five biggest AI hyperscalers – Amazon, Alphabet, Meta, Microsoft, and Oracle – were some of the most aggressive buyback machines on the planet. |
Between 2017 and 2022, this group put an average of 27% of their total cash spending toward buying back their own stock. |
That means more than a quarter of their cash spending went toward repurchases, shrinking the share count and increasing each remaining shareholder’s claim on the business. |
But that deal is starting to change. And Goldman Sachs just showed us why. |
Recently, the investment bank published a research note every investor holding AI mega-cap stocks should pay attention to. |
The five hyperscalers are expected to spend a combined $755 billion on capital projects this year, much of it tied to AI. That’s an 83% increase in just one year. |
Think about that. These companies aren’t simply spending more on AI. They’re nearly doubling down. And that money has to come from somewhere. |
Goldman estimates that spending will amount to roughly 100% of their combined cash flow from operations this year. |
As a group, they’re projected to spend roughly as much on capital projects as they generate in operating cash flow. That leaves very little room for buybacks without tapping cash reserves or taking on more debt. |
Goldman already found what gave first: buybacks. |
In the first quarter of 2026 alone, buybacks among these five hyperscalers fell 64% compared to a year earlier. Their share of total cash spending devoted to buybacks dropped from that old 27% average down to just 15%. |
Friends, when the five companies spending more on AI than anyone else on Earth start pulling back one of the drivers that have supported per-share growth and stock prices for years – it tells you they’re being forced to prioritize the AI buildout over the buybacks shareholders had grown accustomed to. |
|
| |
| |
Big T’s $5 Million AI Bet |
|
|
Big T is going all in on what he believes will be the hottest trend in 2026. |
With this strategy… |
He believes you’ll have the chance to capture massive gains while protecting your money against any AI bubble risk. |
He’s so confident, he’s put over $5 million of his own money into it. |
|
|
|
|
|
| |
|
|
What This Means for Your Money |
If capital spending consumes nearly every dollar these companies generate from operations, very little free cash flow remains for buybacks. |
They can keep repurchasing shares, but only by slowing the buildout, drawing down cash, or taking on more debt. |
Don’t get me wrong. Capital spending can still pay off. If these data centers produce enormous profits, shareholders win over the long run. |
But right now, investors are paying premium prices for these stocks even as the companies scale back their buybacks — betting that the most expensive infrastructure buildout in history will deliver equally extraordinary returns. |
That’s a much higher bar to clear. |
I’ve been warning you about this shift for months. In December 2025, I showed you how AI infrastructure spending was colliding with historically expensive valuations. |
And in January of this year, I showed you roughly $500 billion in market value disappearing from some of the market’s most popular AI names. |
Now Goldman has given us a third warning. The cash that once supported buybacks is being redirected into the buildout instead. |
The Companies Getting the AI Trend Right |
This shift away from buybacks is exactly why I’ve been steering you away from the companies building AI infrastructure and toward the companies deploying it. |
Take Walmart, for example. It’s one of the portfolio positions in my flagship newsletter, The Asymmetric Edge. |
In fiscal year 2026, Walmart generated $14.9 billion in free cash flow and repurchased $8.1 billion of stock, nearly twice as much as the year before. Then in February, its board approved a new $30 billion buyback authorization, the largest in the company’s 55-year history as a public company. |
At the same time, Walmart is using AI and automation to improve supply-chain efficiency, e-commerce economics, and operating margins. |
Walmart isn’t spending hundreds of billions trying to build the next AI model or 650-megawatt data center. It’s plugging AI into a business it already runs. Then it’s turning around and handing more of that cash back to shareholders like you. |
That’s my whole AI thesis in one paragraph. Own the deployers not the builders. These companies don’t need to build AI infrastructure from scratch. |
They simply plug it into their massive, already-profitable businesses and use AI as a profit multiplier. |
Walmart isn’t the only one doing this. |
In my Asymmetric Edge portfolio, we own six other AI deployers moving in the opposite direction from the hyperscalers. They’re generating strong cash flow and increasing the amount they return to shareholders. |
One is a consulting giant that increased its 2026 repurchase plan by $2 billion in June. That brought its total planned buybacks to $7.5 billion, a 62% increase over the year before. |
Another is an energy supermajor that generated $17.2 billion in free cash flow last quarter and used $5.1 billion of it to repurchase shares. |
Out of fairness to my paying subscribers, I can’t give out their names here. But if you’re not yet a subscriber, I encourage you to watch this briefing I recently recorded. In it, I discuss the deployers turning AI into growing earnings, expanding cash flow, and larger shareholder payouts. |
The hyperscalers are telling you — with their own cash — that the easy money in the AI buildout has already been made. The next fortunes will belong to the companies that can turn all that AI spending into real profits for shareholders. |
Let the Game Come to You! |
Big T |
|
|
|
|
|
