Given how ubiquitous the hyper-scaler capex debate is in 2026, it’s probably dawned on a majority of investors that profligate spending on the AI buildout’s bound to come at the expense of buybacks.
I’ve been over this ad nauseam, most recently in “Big Tech Bets Its Balance Sheet On History’s Most Expensive Moonshot.”
The bottom line’s straightforward: If you’re determined to spend every spare dollar on building out, acquiring or accessing compute to stay ahead in the AI arms race, that’s less cash available for buying back stock, all else equal.
When the C-suite goes on a diet vis-à-vis its own cooking, a reliable bid disappears from beneath the stock and the “float-shrink” dynamic which can “artificially” bolster EPS is commensurately diminished.
One way around this is for firms with big AI spending needs to fund capex with debt issuance, thus preserving cash for share repurchases, but that’s not a free lunch: Management says it’s borrowing to invest in the business, critics might say management’s leveraging the balance sheet to keep up buyback appearances. Money’s fungible. You say tomato, I say tomahto.
With that in mind — and in the interest of panning out to a 30,000-foot view on a debate which is almost exclusively centered around just a handful of companies — it’s worth highlighting the figure below, from Goldman’s Ben Snider.
Again, that’s the big picture: Net corporate demand for US equities as the simple difference between gross buybacks and issuance (inclusive of IPOs, follow-ons and converts).
As you can see, the biggest source of demand for US stocks will still be in the market in a big way: Net demand should remain higher than any pre-pandemic year.
“Corporate equity demand should easily outweigh supply [in 2026], although net demand will likely decline relative to the last few years,” Snider remarked, editorializing around the chart.
Goldman expects gross buybacks to top $1 trillion, even if buyback growth stalls due to what Snider described as “shifting management priorities towards capex,” an oblique reference to the hyper-scalers.
Based on companies that’ve reported December-quarter results, S&P 500 buybacks fell 9% YoY last quarter. They rose 5% for the full-year 2025.
Good points. On a smaller scale some extra sales growth which raises cash flow other things equal, can help.That will probably be a bigger factor 2-3 years from now. The model for large cap tech has changed at least in the short to intermediate term. No longer asset light and low balance sheet leverage. In my eyes this should reduce/eliminate returns from increasing valuations. Returns will be far more dependent for large cap tech from profit growth.
I never underestimate the C-Suite’s greed, which can lead to financial engineering (including buybacks), if necessary, in order to meet EPS targets in their executive compensation plans.
Curious = do you think that their “versus issuance” offset to versus Gross buybacks their way of calculating NET buybacks which strips out buybacks to offset options and restricted stock issued to employees?
For investors, the net buyback number is more important in the short run
Should buybacks even be allowed? Weren’t they illegal until the “greed is good” 80s?
Buybacks have been around since well before then. They used to be packaged as tender offers. Here is my all time favorite buyback story.
One of my individual stock holdings has been executing a modern day version of this story for the past decade. I was actually present at the annual meeting (via zoom) when a representative of a large Canadian pension fund handed the CEO a copy of this story in about 2013.
Still invested in this stock.
https://novelinvestor.com/the-teledyne-buyback-effect/