Nvidia's $150 Billion Buyback Ceiling, Explained
Nvidia added $150 billion to its existing stock repurchase plan. Reuters The figure was reported Sept. 28, 2026.
The September capacity follows a May authorization. Nvidia's board authorized an additional $80 billion for repurchases in May, when $38.5 billion remained under the prior program at the end of the quarter. Yahoo Finance
That May top-up followed an earlier program. In August 2023, Nvidia authorized a $25 billion buyback with no expiration date. The Motley Fool
The broader context here is how large-company buyback programs are built. An authorization is a ceiling, like a credit limit. It does not schedule spending. The open-ended wording used in 2023 keeps timing flexible around earnings blackouts, choppy markets and cash conversion cycles. Adding new capacity before old capacity runs out avoids a gap in that flexibility. The steps from $25 billion to an extra $80 billion to an extra $150 billion fit that pattern. New headroom arrives while old headroom is still unused.
In my view, investors tracking the math should separate permission from action. Permission sets the maximum. Action decides how much the tradable share count actually falls, what average price the company pays, and when any lift to earnings per share shows up. That is why professionals watch daily buying pace versus trading volume, preset Rule 10b5-1 buying plans, and possible accelerated share repurchase deals that retire shares quickly. What happens afterward also matters. Cancelled shares are gone for good. Shares held as treasury stock can be reissued later for employee pay.
Looking at what this means for capital allocation, buybacks compete with research and development, capital spending on equipment and plants, and cash cushions. They also compete with dividends as a way to return cash. Buybacks give management choice on timing and can lift earnings per share by shrinking the share count. Dividends give regular, visible income. A board that keeps adding buyback capacity is choosing flexibility and per-share math over a fixed payout. That choice reveals tolerance for uneven payouts. It does not reveal a forecast.
In my view, valuation needs to separate the message from the money. The message is willingness to use large amounts of cash to retire stock. The money only moves when shares are actually bought. Until then, enterprise value, the total value of the business, should not assume fewer shares. New employee stock grants, forfeitures and tax withholding can offset some buying. What matters is the net change in share count, not the headline limit. That is why trading desks track reported weighted-average diluted shares each quarter rather than projecting from board limits.
Looking at what this means for governance, repeated large authorizations put more timing power with managers. That is efficient when the stock trades easily. It also makes clear disclosure of pace and average price paid more important. For savers holding the stock in a pension or index fund, those two execution details will tell you more than the $150 billion figure itself.
