What 15 Years of Stock Buybacks Really Did
Buybacks shrink the share count far less than the headline spending suggests. Across 411 stocks in this site's 15-year data, companies spent about $8.5 trillion on repurchases, yet the median share count fell only 10%, and 148 of the 411 companies have more shares than 15 years ago.
- Stock-based pay offset about $19 of every $100 spent, and mergers paid in shares undid more.
- The winners (AutoZone, O'Reilly, Ameriprise, Apple) shrank the count by 40% to 60%; the biggest spenders on this list are not all the biggest shrinkers.
This is educational analysis, not investment advice.
How much have companies spent on buybacks, and what did they get?
Total buyback spending over 15 years across 411 stocks with full data. The median company still shows only 10% fewer shares, and 148 have more.
| Measure | Figure |
|---|---|
| Stocks with a full 15-year record | 411 |
| Total spent on buybacks | About $8.5 trillion |
| Total stock-based compensation | About $1.6 trillion, or 19% of buybacks |
| Companies with a lower share count than 15 years ago | 263 of 411 |
| Median change in share count | -9.9% |
| Companies whose stock pay exceeded their buybacks | 38 of 394 that bought back stock |
Share counts here are restated for splits by chaining values across annual reports, so they measure real ownership dilution. The data covers the 15 fiscal years to 2025. Dividends vs buybacks explains why buybacks became the main way to return cash.
Which companies spent the most, and did their share counts fall?
| Stock | Spent on buybacks | As a share of today's market value | Change in share count |
|---|---|---|---|
| AAPL | $816 billion | 16% | -43% |
| GOOGL | $348 billion | 9% | -7% |
| MSFT | $253 billion | 7% | -12% |
| WFC | $166 billion | 67% | -39% |
| JPM | $162 billion | 18% | -29% |
| ORCL | $156 billion | 41% | -43% |
| XOM | $153 billion | 21% | -12% |
| BAC | $150 billion | 43% | -25% |
| V | $113 billion | 15% | -30% |
| CSCO | $102 billion | 24% | -26% |
Percentage fall in share count for the ten biggest buyback spenders (green: 20% or more). Alphabet and NVIDIA barely moved because stock pay and issuance offset the buybacks.
Apple is the outlier, with the largest spending and a 43% smaller share count. Alphabet spent about $348 billion and shrank its share count by only 7%, because stock pay absorbed about half of that spending.
Which companies shrank their share counts the most?
| Stock | Change in share count | Spent on buybacks | As a share of today's market value |
|---|---|---|---|
| AIG | -68% | $72 billion | 168% |
| EBAY | -64% | $48 billion | 88% |
| AZO | -61% | $30 billion | 57% |
| AMP | -60% | $27 billion | 63% |
| ORLY | -58% | $27 billion | 37% |
| MPC | -57% | $56 billion | 43% |
| LOW | -56% | $76 billion | 68% |
| HPQ | -55% | $43 billion | 144% |
| HCA | -52% | $49 billion | 48% |
| MCK | -51% | $32 billion | 27% |
The list is led by companies with steady cash flow and a habit of repurchasing for decades: AutoZone and O'Reilly, Ameriprise, eBay, Allstate, Aflac, Apple. AIG's large fall includes its post-2008 restructuring, so treat it as a special case. For owners, the effect is a bigger share of the same earnings: a 40% fall in the count lifts profit per share by about two thirds with no growth at all.
Which companies spent heavily but did not shrink?
| Stock | Spent on buybacks | As a share of today's market value | Change in share count | Likely reason |
|---|---|---|---|---|
| CNC | $10 billion | 27% | +135% | Issued shares in Health Net and WellCare deals |
| TMUS | $58 billion | 36% | +111% | Issued shares in the Sprint merger |
| FIS | $14 billion | 73% | +71% | Issued shares to buy Worldpay |
| BMY | $35 billion | 28% | +19% | Issued shares to buy Celgene |
| T | $43 billion | 30% | +16% | Issued shares in the DirecTV and Time Warner deals |
| CI | $41 billion | 50% | -2% | Issued shares to buy Express Scripts |
| FISV | $9 billion | 34% | -5% | Issued shares in the First Data merger |
The reasons in the last column are the standard explanations, not figures from this data: when a company pays for a big acquisition with its own shares, those buybacks end up undoing the dilution instead of shrinking the count. Check each company's history before relying on them. The pattern is that buyback spending counts less than the net change in shares.
What about Oracle, which spent $156 billion on buybacks?
| Measure | Oracle (ORCL) |
|---|---|
| Spent on buybacks in 15 years | $156 billion |
| Change in share count | -43% |
| Today's market value | $379 billion |
| Debt now | More than $160 billion, per Bloomberg |
Oracle retired about 43% of its shares at a time when the stock was far lower, and is now borrowing more than $160 billion to build AI data centres at a share price about half of its high. That is the weakness in buybacks: the cash leaves, and later the company must borrow at a worse moment to fund growth. Why five profitable stocks are down about 50% covers Oracle's debt and the other fallen names, and will Western Union cut its dividend? shows what happens when buybacks are paused to protect a payout.
How can you check whether a company's buybacks are working?
- Look at the share count, not the spending. Compare shares outstanding today with five or ten years ago on the company's financial pages.
- Subtract stock-based pay. A buyback that only offsets stock compensation is a cost of doing business, not a return of cash. This site's Shareholder Yield Leaders page nets it out.
- Check what price it paid. Buying at a peak can destroy value: Oracle shows how heavy repurchases can precede a fall.
- Check the funding. Buybacks funded by free cash flow are healthy, those funded by debt are riskier.
- Watch mergers. A big share-funded deal can erase a decade of repurchases.
Sources
- Shareholder Yield Leaders on this site: 15-year buybacks, dividends, stock pay and share counts
- Method: shareholder yield: how share counts are restated for splits
- Yahoo Finance: Oracle stock falls on more debt to fund AI chip buying (October 8, 2026): Oracle debt
Buybacks, dividends, stock compensation and share counts are from SEC filings as processed by this site's weekly refresh, for 411 stocks with a complete 15-year record to fiscal 2025 (companies with a shorter record, such as spin-offs, are excluded). Totals add up each company's yearly figures and are not adjusted for inflation. The reasons for share-count increases are general explanations of well-known deals, not findings of this data. This is educational content, not investment advice.
Stock buyback FAQ
Do stock buybacks really reduce the share count?
Less than headlines suggest. Across 411 stocks, companies spent about $8.5 trillion in 15 years, but the median share count fell only 10%, and 148 companies have more shares.
Which companies shrank their share count the most?
AutoZone, O'Reilly, Ameriprise, eBay, Allstate, Aflac and Apple cut their counts by more than 40% over 15 years, with AIG a special case.
Why do some companies buy back stock and still have more shares?
Stock-based pay and acquisitions paid in shares issue new shares, which can offset or exceed what buybacks retire.
Are buybacks better than dividends?
Neither is better in general. Buybacks help owners only if the share count really falls and the price paid is reasonable; dividends are paid in cash regardless.
How much did Apple spend on buybacks?
About $816 billion over 15 years, and its share count is about 43% lower.