How important are buybacks?

That’s a rhetorical question. The answer, qualitatively, is “very.”

If you want the numbers — i.e., a quantitative answer — the corporate bid equates to something like $5 billion per day in US equity demand. Authorizations in the US are north of $1.3 trillion in 2025 (looking across the Russell 3000), and executions will almost surely hit a record before the year’s out.

One of many big questions swirling around massive hyper-scaler capex plans asks how hundreds of billions in AI-related spending will impact those companies’ repurchase programs.

Even if you assume ~half of AI outlays will be funded with new borrowing, that leaves a lot to free cash flow and, presumably, less for buybacks.

The figures above, from Nomura’s Charlie McElligott, illustrate a lot of key points.

First, overall authorizations (and, ultimately, executions) won’t necessarily fall just because some of the biggest names repurchase fewer shares as they divert cash to the AI buildout.

Second, the biggest names are still buying back a lot of stock. Management may be lost irretrievably down the AI rabbit hole, but corporate America’s addicted to buybacks. 40 years on from the unofficial inauguration of shareholder capitalism, buybacks are as American as apple pie. So, we probably shouldn’t fret too much about this.

Third, we should maybe fret a little. Because as the table beside the chart shows, companies engaged in the AI arms race are set to account for a much lower share of overall S&P 500 buybacks than they did in 2024, and if that trend continues in 2026, it’ll be felt.

“The pace of spending plans is beginning to outstrip FCF, and thus poses significant implications for the overall corporate [bid] which has been the primary source of equities demand and de facto vol supply over the past decade,” McElligott said, noting that at the very least, we’re staring down a negative “flow over stock phase shift” for buybacks in 2026.