The Buyback Floor: How $1.21 Trillion in Share Repurchases Became Wall Street’s Biggest Magic Trick

The Magnificent 7 didn’t grow their way to higher earnings. They shrank their share counts. And now AI capex is eating the cash that made it possible.

A Brief History of Legalized Price Manipulation

Here’s something most investors don’t know, and it’s kind of wild when you think about it: before 1982, if a company bought its own shares on the open market to push the price up, that was market manipulation. Full stop. The SEC could come after you.

Then, under the Reagan administration, the SEC adopted Rule 10b-18. It didn’t legalize buybacks exactly — it created a “safe harbor.” As long as companies followed four conditions (one broker per day, no opening or closing trades, price limits, volume limits), they could repurchase shares without the SEC treating it as manipulation.

The result? In 1982, S&P 500 companies spent about $2 billion on buybacks. By 2024, that number crossed $900 billion. In 2025, the trailing twelve months hit $1.02 trillion.

Let that sink in. The single largest source of demand for US equities — bigger than pension funds, bigger than 401(k) contributions, bigger than foreign investors — is companies buying their own stock. And it was illegal 44 years ago.

Apple: The Greatest EPS Illusion in Market History

Let’s talk about Apple, because it’s the cleanest example of what buybacks actually do when deployed at massive scale.

Apple’s net income over the past five years: approximately $95 billion, $100 billion, $97 billion, $94 billion, $101 billion. Flat. Essentially sideways. The iPhone is a mature product. Services revenue grows, but hardware margins compress. The company is printing roughly the same amount of profit year after year.

Apple’s earnings per share over that same period? Up and to the right. Reliably. Every quarter.

How? They retired approximately 15% of their outstanding shares. When you earn $100 billion and divide it by 15% fewer shares, EPS jumps like you grew profits by 18%. You didn’t. You just bought back stock.

Roughly 85% of Apple’s EPS growth over this period came from share retirement, not profit growth.

This isn’t a secret. It’s in the filings. But here’s what’s strange: most of Wall Street’s valuation models are built on EPS. Price targets, forward PE ratios, growth projections — they all use earnings per share. So a company that masters the buyback machine can deliver “EPS growth” indefinitely without ever growing the actual business.

Until it runs out of cash to buy. Or finds something else to spend it on.

The SBC Shell Game

If Apple is the cleanest example of buyback-as-EPS-engine, Meta and NVIDIA show you the other trick: buyback-as-hidden-payroll.

Stock-based compensation (SBC) is how Big Tech pays its talent. Engineers, executives, middle managers — a significant chunk of their comp comes in shares. Every quarter, new shares are issued to employees. This dilutes existing shareholders. More shares outstanding, same earnings, lower EPS.

So what do you do? You buy back shares to offset the dilution.

Meta dedicates approximately 90% of its buyback spend to neutralizing stock-based compensation. They’re not “returning cash to shareholders.” They’re running in place. The buyback is paying the payroll bill — it just doesn’t show up on the income statement as an expense.

NVIDIA is even more striking. They spent $21 billion on buybacks. Shares outstanding dropped by 0.6%. Where’d the other $20.8 billion go? Into absorbing the dilution from stock-based compensation. The buyback looked like a shareholder return. It was actually payroll laundered through the balance sheet.

The Mag 7 Buyback Scoreboard

Here’s where the money went over the past five years:

Notice the spread. Apple is using buybacks as a profit multiplier. Meta is using them as hidden payroll. NVIDIA is treading water. Amazon and Tesla are net diluters — their SBC issuance outpaces any repurchases.

Combined: $1.21 trillion. Roughly the GDP of Mexico. Spent not on new products, new factories, new employees, or new R&D — but on buying shares to make the earnings math work.

The Insider Selling Correlation

Here’s where it gets uncomfortable.

There is a 0.85 correlation between the timing of corporate buyback execution and insider selling windows. That’s not 0.85 out of 100. That’s 0.85 out of 1.0. For reference, anything above 0.7 is considered a strong statistical relationship.

What this means in plain English: when companies are actively buying back shares (pushing the price up or putting a floor under it), insiders are simultaneously selling their personal holdings into that artificial demand.

The company announces a $50 billion buyback. The stock pops. Over the next 90 days, the company executes purchases that support the price. During those same 90 days, the CEO, CFO, and VP of Engineering sell shares worth $200 million into the company’s own bid.

This is legal. The insider selling is pre-scheduled under 10b5-1 plans. The buyback is conducted within 10b-18 safe harbor rules. Both are disclosed. Neither is hidden. But the combined effect is that corporate cash — which belongs to all shareholders — is being used to provide liquidity for insider sales at supported prices.

Ask yourself: if you ran a lemonade stand and used the company bank account to buy your own cups while simultaneously selling your personal cups at the inflated price, would anyone call that “returning value to shareholders”?

Buyback Peaks Predict Market Tops

This is the part that matters most for what happens next.

Historical pattern, and it’s remarkably consistent:

The pattern: when buybacks hit a crescendo, it often means companies have run out of better things to do with their cash, multiples are stretched, and the easy financial engineering gains are priced in. The peak in buyback activity doesn’t cause the correction — but it consistently signals that the cycle is mature.

The Current Signal: The Floor Is Cracking

Trailing 12-month aggregate buybacks peaked in September 2025 at $1.02 trillion. Since then, growth has flattened.

Why? One word: capex.

The same Mag 7 companies that were spending a combined $240 billion per year on buybacks are now spending $250–300 billion per year on AI infrastructure. Data centers. Custom chips. Cooling systems. Power contracts. The AI arms race requires real capital investment — the kind that actually builds things — and it’s competing directly with the cash that used to fund buybacks.

Microsoft’s capital expenditure doubled in two years. Meta’s nearly tripled. Alphabet is spending more on data centers than it spent on buybacks for the first time. Even Apple, the buyback king, is signaling increased AI infrastructure spending.

Something has to give. You can’t spend the same dollar on a GPU cluster and on buying back shares. The companies that built their EPS story on financial engineering now face a choice: keep the buyback floor intact, or invest in AI and hope the actual business growth makes up for the loss of share retirement.

What This Means for Your Portfolio

If you own the Mag 7 — and if you own an index fund, you do — you need to understand what’s been holding these stocks up.

It wasn’t just earnings growth. In many cases, it wasn’t earnings growth at all. It was a $1.21 trillion campaign to make the EPS number go up by making the share count go down. That campaign is now competing for capital against the biggest infrastructure buildout since the transcontinental railroad.

Our engine tracks these dynamics alongside 167 cross-industry cascade effects. Because buyback trends don’t exist in isolation — they interact with AI adoption velocity, capital expenditure cycles, and the structural forces reshaping every sector.

The floor was always artificial. Now it’s competing with reality.

Related reports: The Autopilot Bid  |  The Trump Overlay  |  The Sideways Layoffs  |  The Jevons Paradox