A minimalist cover on a near-black canvas: a vertical price bar rising toward a bright horizontal ceiling line labelled 'intrinsic value', with the zone above the line shaded as a do-not-buy region. A small 'Ask Warren' kicker sits above the headline 'Should we buy back stock at this multiple?' in white sentence-case type.

Ask Warren: Should we buy back stock at this multiple?

The question CFOs are asked at almost every Q4 board meeting now is some version of: what are we doing about the buyback? Authorisations are open, balance sheets are full, the AI capex cycle is forcing every dollar of free cash flow into a conscious choice, and activists are watching. The pressure is to do something — announce, accelerate, expand, or at minimum reaffirm.

Buffett’s writing on buybacks gives the CFO a hard answer that most boardroom conversations skip over. Two letters — 2011 and 2016 — together form the framework.

The framework (2011)

Charlie and I favor repurchases when two conditions are met: first, a company has ample funds to take care of the operational and liquidity needs of its business; second, its stock is selling at a material discount to the company’s intrinsic business value, conservatively calculated.

— Buffett, 2011 letter

Two conditions. Both are gates. Either one missing and the buyback should not happen — not be paused, not deferred, not staggered. The same letter closes the loop with the line that should sit on the boardroom whiteboard: what is smart at one price is dumb at another.

That reframes the buyback authorisation itself. An open authorisation with no price ceiling is not a capital allocation decision. It is a delegation of capital allocation to whatever the share price happens to be on the day buying executes. The CFO who wrote the case is no longer the one making the call; the trading desk is.

The math (2016)

Five years later Buffett restated the principle without the corporate vocabulary:

If there are three equal partners in a business worth $3,000 and one is bought out by the partnership for $900, each of the remaining partners realizes an immediate gain of $50. If the exiting partner is paid $1,100, however, the continuing partners each suffer a loss of $50.

— Buffett, 2016 letter

This is the test stripped of all financial-press framing. A buyback is a partial buyout of one shareholder by the remaining shareholders. The remaining shareholders win if they pay less than intrinsic value per share, and lose if they pay more. There is no “the buyback supported the share price” benefit to the continuing owners — only the dollar gap between price paid and value retained, multiplied by shares bought.

Buffett’s pointed observation:

It is puzzling, therefore, that corporate repurchase announcements almost never refer to a price above which repurchases will be eschewed.

— Buffett, 2016 letter

In other words: every buyback announcement should include the price above which the buyback will not be made. The fact that most don’t tells you something about whose interests the authorisation is actually serving.

What this means for a 2026 CFO

Three translations.

  • Treat every buyback authorisation as a price-conditional decision, and put the price in writing. A board approving a $500M repurchase without naming the intrinsic value or the ceiling above which it should not be made is approving an open-ended currency conversion, not a capital deployment. The intrinsic value calculation is the work; the dollar amount is just the budget for it.
  • Apply the two-gate test honestly before every tranche. Are the operating reserves genuinely sufficient — including a downside case where the AI capex cycle keeps surprising, demand softens, and refinancing costs rise? Is the share price still below your conservatively-calculated intrinsic value at today’s multiple, not the one that drove the original authorisation? Either gate failing means pause, not “wait for the next window.”
  • Be willing to say no in public. The hardest version of this is the active not-buying decision. If the stock has run, the framework says the buyback stops — and the CFO has to defend that publicly, against investors who have priced in the buyback flow. The discipline here is the discipline of being willing to disappoint the buy side because the math does not support it. The framework is only credible if the no is real.

The buyback question looks like a treasury decision. It is not. It is a one-dollar test decision dressed up in a board resolution. The 2011 and 2016 letters are the cleanest writeup of how to think about it that the function has, and they are still ahead of where most boardroom conversations are in 2026.