This is BMInsider PRO content
You're reading a Deep Dive — our most in-depth research. Unlock full access to all analyses, Smart Money breakdowns & more.
Unlock BMInsider PRO →One-time payment · instant access · no account needed
In the first quarter of 2026, Alphabet did not repurchase a single share of its own stock for the first time in 33 quarters. Meta took its buyback line to zero. Microsoft repurchased at close to a decade low. And in that very same quarter, S&P 500 companies collectively bought back more of their own stock than in any first quarter on record. Both statements are true — and what happened in the space between them is reshaping the structure of the U.S. equity market more than any rate decision this year.
For more than a decade, buybacks have been the single most dependable source of demand for American shares. Not mutual funds, not retail, not foreign investors: companies themselves were the largest net buyer. In 2026 that buyer did not leave. It changed its name. The cash that used to flow from Alphabet’s balance sheet into Alphabet’s stock now flows into data centers, and someone else is filling the gap. Who that someone is, and why it matters more to a portfolio than the headline total, is what this analysis is about.
The biggest buyer in the market appears in no fund flow report
Start with the scale, because it is routinely underestimated. In the twelve months through September 2025, S&P 500 companies spent $1.020 trillion repurchasing their own shares, according to S&P Dow Jones Indices — up 11.1 percent year over year. Over the same period, those same companies paid $664.9 billion in dividends. Buybacks are not the junior partner to the dividend. They are roughly one and a half times its size.
A company buying its own stock is also a buyer of unusual quality. It is price-insensitive, it executes on a program rather than on sentiment, and it typically keeps buying when others are selling. That is precisely why the phrase buyer of last resort attached itself to repurchase programs in the first place.
The distribution of that trillion, however, is extremely lopsided. The twenty largest repurchasers accounted for roughly 49.5 percent of all buybacks in the third quarter of 2025 — and 51.3 percent in the quarter before that, against a historical average of 47.8 percent and 44.5 percent pre-pandemic. Half of the largest buyer in the U.S. equity market consists of twenty balance sheets. When five of them change direction simultaneously, that is not a footnote.
2026: a record and a standstill in the same quarter
That is exactly what happened, and the year’s headlines appear to contradict each other. Resolving the contradiction is the whole point.
On one side: in the first four months of the year, S&P 500 companies announced $665 billion in repurchase authorizations, according to Birinyi Associates — the strongest start to a year ever recorded. Executed net buybacks in the first quarter reached roughly $270 billion by Deutsche Bank’s count, about $300 billion gross, also a record.
On the other side: Goldman Sachs expects gross buyback growth of just 3 percent for full-year 2026, against 33 percent growth in capital expenditure across the same index. One first-quarter tally put it even more starkly: capex up 38 percent, buybacks up 1 percent. S&P 500 capital spending has climbed from roughly $1 trillion annualized to about $1.5 trillion, and two thirds of that increase comes from five companies.
Both descriptions are accurate because they measure different things. The total is holding. The composition has flipped. And an authorization is not an execution: a repurchase program is an option, not an obligation. It can sit unused for years, and in quarters when the balance sheet has something else to fund, it does.
| Metric | Value | Context |
|---|---|---|
| Buybacks, 12 months to 9/2025 | $1.020 trillion | up 11.1% year over year |
| Dividends, 12 months to 9/2025 | $664.9 billion | buybacks are roughly 1.5x dividends |
| Announced Jan–Apr 2026 | $665 billion | strongest start to a year on record |
| Executed net, Q1 2026 | ~$270 billion | quarterly record; ~$300bn gross |
| Expected buyback growth 2026 | +3% (Goldman Sachs) | capex in the same index: +33% |
| S&P 500 capex, annualized | ~$1.0tn → ~$1.5tn | two thirds of the rise from five firms |
The five that stopped
Alphabet is the cleanest case. The company repurchased stock for 33 consecutive quarters — $61.5 billion in 2023, $62.2 billion in 2024, still $45.7 billion in 2025. In the first half of 2026 it spent zero, against $28.3 billion in the first half of 2025. It is the first stretch without a repurchase since late 2017.
The offsetting entry sits on the same cash flow statement. Alphabet’s capital expenditure rose from $27.9 billion in the fourth quarter of 2025 to $35.7 billion in the first quarter of 2026 and $44.9 billion in the second. Full-year guidance was set at up to $185 billion in February and raised in July to a range of $195 billion to $205 billion. Second-quarter free cash flow came in at negative $5.86 billion — the first negative print since the 2004 IPO.
What Alphabet did instead is the telling part. In February 2026 it priced a bond offering upsized from $15 billion to $20 billion against an order book north of $100 billion, then extended the global raise past $30 billion, including roughly $11 billion in sterling and Swiss francs — and, as a signal-rich curiosity, a hundred-year sterling bond, the first century paper from a technology issuer since Motorola in 1997. A company that spent years returning part of every quarter’s cash to its own shareholders is now borrowing on a hundred-year horizon. That is not a tweak to capital allocation. It is a regime change.
Meta also took its repurchase line to zero in the first quarter of 2026 while capex climbed from $12.9 billion to $19.0 billion. Microsoft was the only one of the four largest AI spenders still buying back stock at all in the quarter, at a volume near a multi-year low, with capex up 84 percent. Amazon pays no dividend, repurchases almost nothing, and is planning capital spending on the order of $200 billion for 2026. Together with Oracle, the five carry a combined 2026 capex budget of roughly $755 billion, up 84 percent from 2024; consensus for 2027 sits near $920 billion.
| Company | Buybacks | Capital spending |
|---|---|---|
| Alphabet | $62.2bn (2024) → $45.7bn (2025) → $0 in H1 2026 | 2026 guidance raised to $195–205bn |
| Meta | zero in Q1 2026 | $12.9bn → $19.0bn quarter over quarter |
| Microsoft | reduced, near a multi-year low | quarterly capex up 84% year over year |
| Amazon | no dividend, effectively no repurchases | 2026 plan around $200bn |
| Apple | up to $100bn authorized (April 30, 2026) | about $4.3bn in the first fiscal half |
The other 495 closed the gap
If five companies that accounted for a meaningful share of repurchase volume withdraw at the same time and the aggregate still sets a quarterly record, the rest must have bought a great deal more. The data say exactly that: net repurchases by non-hyperscaler S&P 500 companies rose roughly 30 percent year over year in the first quarter of 2026.
Who those buyers are was already visible in 2025. In the third quarter of that year, financials accounted for 26.2 percent of all buybacks — $65.3 billion, up 26.3 percent from the prior quarter — nearly level with information technology at 28.4 percent. Large banks releasing capital after the stress tests, insurers with heavy excess solvency, energy majors for whom capital discipline became doctrine after the shale bust, consumer staples and healthcare names with predictable cash flows: that is the profile of the new marginal buyer.
What these companies share is a quality that has become scarce: they have cash flow and no project large enough to absorb a hundred billion dollars. For them the buyback is the obvious use of capital. For the hyperscalers it has become the most expensive alternative in 2026 — not because the stock is unattractive, but because in a capacity race every dollar not invested counts as ground conceded.
Why the composition matters more than the total
Here is the core of it, and it is almost universally skipped in the coverage. The S&P 500 is capitalization-weighted. The five companies that stopped repurchasing are among its heaviest constituents. The 495 that are buying more are, on average, far lighter.
Which means share counts will now shrink where index weight is small, and stagnate or grow where index weight is large. At the very top, net supply is turning positive — Alphabet’s first-quarter capital raising exceeded the entire secondary issuance of the rest of the S&P 500 in that quarter. Meanwhile stock-based compensation, unusually heavy at large technology companies, is no longer being neutralized by repurchases. What stayed invisible for years, because the buyback swept the dilution away quarter after quarter, is now visible in the share count.
For investors this has three concrete consequences. First, a mechanical tailwind is shifting from the cap-weighted index toward equal-weighted and value-tilted segments. Second, the claim that buybacks support the market loses its generality: from 2026 they support a different part of the market than they did for the preceding decade. Third, market behavior in drawdowns changes. The price-insensitive buyer that kept bidding through weakness is, for now, absent from the largest names — precisely the names whose valuations lean hardest on momentum.
Apple, the control experiment
Apple shows how much of this is a choice rather than a law of nature. On April 30, 2026, its board authorized the repurchase of up to $100 billion of stock and raised the quarterly dividend 4 percent to 27 cents. Across the entire first half of its fiscal year, Apple spent roughly $4.3 billion on capital expenditure — about a tenth of what Alphabet spent in a single quarter. Since the program began in 2012, Apple has returned more than $1 trillion to shareholders, roughly $850 billion of it through buybacks.
Apple has chosen the mirror image of the hyperscaler strategy: return on capital employed over optionality. You can read that as discipline or as an admission that the company has run out of ideas large enough to absorb the cash. Both readings are defensible, and that is exactly what makes the case useful. It shows that capital allocation in this cycle is a genuine fork in the road, not a forced adjustment. Five years from now we will know which path produced more value per dollar. Nobody knows today, and anyone claiming otherwise is selling an opinion as analysis.
What a buyback actually delivers — and what it does not
Because the topic carries heavy ideological freight, a sober decomposition is worth the space. A buyback creates no value by itself. It divides an unchanged enterprise value across fewer shares. It is value-accretive only when the company buys below intrinsic value; above it, the transaction transfers value from continuing holders to selling ones.
Which names the central problem: companies buy pro-cyclically. Repurchase volumes peak in years of high valuations and collapse in crises — in 2020 programs were suspended en masse, exactly when the shares were cheap. The moment when the most cash is available for buybacks is systematically the moment when buybacks are worth the least.
Equally important: buyback dollars are not share count reduction. A substantial portion of most programs merely offsets dilution from stock-based compensation. The number that matters is therefore not the dollar figure but the trajectory of shares outstanding. Here the S&P Dow Jones data show how small the group of genuine shrinkers is: only 17.1 percent of index members reduced their share count by at least 4 percent year over year in the third quarter of 2025, up from 13.6 percent a year earlier. For more than four fifths of the index, the buyback accomplishes less than its headline dollar amount implies.
The 1 percent excise tax on net repurchases, in force since 2023, has changed little. It reduced S&P 500 operating earnings by 0.36 percent in the third quarter of 2025 and by 0.40 percent on a twelve-month basis. As a policy lever that is negligible; index-provider modeling suggests a rate of 2 percent or higher would be needed before it visibly bit into volumes and share count reduction. An increase is not currently on the legislative agenda, but it belongs on every risk list.
| Claim | How well it holds up |
|---|---|
| Buybacks lift earnings per share | Mechanically true, but only where the share count actually falls — for 82.9% of index members it fell by less than 4% year over year |
| Buybacks create value | Only below intrinsic value; above it, a transfer to the selling shareholder |
| Buybacks support the market | True for the past decade, but from 2026 for a different segment of it |
| Buybacks come at the expense of investment | Demonstrable for the first time in 2026 — in the opposite direction from the long-standing claim |
| The excise tax restrains programs | Not measurably at 1% (0.36% earnings impact); relevant from roughly 2% |
Tax, the quiet part of the return
For U.S. taxable investors, the rotation has a second-order effect that rarely makes the coverage. A dividend is taxable in the year it is received, whether the investor wants the cash or not. Qualified dividends carry preferential rates, but the timing is not optional. A buyback is not a taxable event for the continuing shareholder at all: the value shows up in the share price and is taxed only on sale, at the holder’s choosing, and at long-term rates if the position has been held beyond a year. At identical pre-tax returns, the repurchase is structurally the more tax-efficient form of capital return in a taxable account — the deferred tax keeps compounding.
That advantage argues for holding the higher-yielding, dividend-heavy names in tax-advantaged accounts and letting buyback-driven compounders sit in the taxable one, not the reverse. It also means the rotation described above quietly changes the after-tax profile of a portfolio: the 495 doing the repurchasing skew toward financials, energy and staples, which are also the highest dividend payers in the index. An investor who follows the buyback bid into those sectors picks up more current taxable income along with it.
One more wrinkle deserves attention. Because the excise tax applies to net repurchases, issuance offsets it — which slightly favors companies that repurchase while also issuing shares to employees. It is a small effect at a 1 percent rate. It would not be small at 4 percent, a level that has surfaced in past legislative proposals and would turn a rounding error into a genuine allocation input.
Three scenarios into 2027
Where does this go? Three paths are plausible, and they differ less in the aggregate total than in its distribution.
| Scenario | Path | Portfolio consequence |
|---|---|---|
| Rotation persists (base case) | Hyperscalers stay investment-led, 2027 capex consensus near $920bn; financials, energy and healthcare carry the buyback volume | Share count shrinks in the belly of the index; equal-weighted and dividend-adjacent segments gain a mechanical edge |
| The capex cycle turns | Projects get stretched, depreciation schedules debated, cash flow returns; repurchases restart at the top of the index | A powerful tailwind for the heaviest names; timing hinges on utilization data, not on announcements |
| Both engines stall | An earnings downturn hits the 495 as hard as the 5; programs are suspended, as in 2020, exactly as prices fall | Losing the price-insensitive buyer amplifies drawdowns; historically the real risk |
The third scenario deserves the most attention precisely because it gets the least. Buybacks are not a constant but a residual: they are funded from what remains after capital spending, interest and dividends. In an earnings downturn the repurchase is the first line cut — and because it is the largest net buyer, that demand disappears exactly when it is most needed. This pro-cyclicality is the built-in design flaw of the system, regardless of who happens to be doing the buying.
The strongest counterargument
It would be dishonest to close without the other side. Deutsche Bank concluded in late June 2026 that the capital spending surge is unlikely to undermine buybacks across the broader U.S. equity market. Rising inflows into U.S. equities, elevated household cash balances and continued earnings growth should absorb the shortfall, even against higher share issuance. And the numbers support that view: a quarterly record is a quarterly record, no matter who delivers it.
Follow that logic and the conclusion is calmer: the market absorbed the withdrawal of its five largest buyers within two quarters without aggregate demand breaking — evidence of depth, not fragility. The second counterargument is equally serious. Investment in productive capacity is the economically higher use of capital. An economy in which corporations build rather than retire shares is not the worse one. For shareholders that holds only if the return on the invested capital ultimately clears the cost of capital — and on this cycle, that question is genuinely open.
What to watch
Four observable measures say more than the headline total. First, shares outstanding on the balance sheet, not the repurchase dollars in the press release; check across four to eight quarters whether the count is actually falling. Second, the ratio of buybacks to stock-based compensation — below one, the program is pure dilution defense. Third, at the large technology names, the language around data-center depreciation schedules, which determines how quickly today’s capex becomes tomorrow’s earnings and when repurchases can resume. Fourth, the sector split of buybacks in the index provider’s quarterly reports: if the financials share keeps climbing past technology, the rotation is confirmed.
None of that is a recommendation. It is a clarification. Anyone who held a cap-weighted U.S. index over the past decade received the buyback tailwind for free, because it blew where the weight was. That coupling broke in 2026. The effect can still be harvested — it simply has to be sought more deliberately than before.
Bottom line
The story that the largest buyer of American equities is retreating is wrong. The story that nothing has changed is equally wrong. What is true: the trillion is still there, but it comes from different balance sheets. Five companies that supplied a substantial share of the volume have redirected their capital into data centers; the rest lifted their repurchases by roughly 30 percent and closed the gap.
That frames the decisive question for the next two years. It is not whether buybacks come back — they never left. It is whether an investment wave running at $1.5 trillion a year ultimately generates more earnings per share than the share-count reduction it displaced. Until that is settled, the unglamorous finding stands: the mechanical tailwind that carried the heaviest index positions for a decade is now blowing somewhere else. Anyone still positioned where it used to blow should know it is no longer at their back.