Nvidia’s $150 Billion Buyback Tests Excess Cash

Nvidia’s board added $150 billion to its share-repurchase authorization on September 28, lifting total remaining capacity to $235 billion through fiscal 2028. Reuters described it as the largest authorization on record. The number is enormous, but the useful finance question is not whether Nvidia can announce it. It is how a board decides that cash is genuinely excess while the company is still funding one of the largest technology investment cycles in history.

The judgment

• Implication: Treat the authorization as flexible capacity, not committed spending. Its value depends on whether repurchases occur only after funding strategic investment, liquidity, and downside resilience.
• What would change the conclusion: Sustained free cash flow, durable margins, lower reinvestment needs, and consistent net share-count reduction would strengthen the case. Slower AI demand, higher capital commitments, or repurchases that mainly offset dilution would weaken it.
• Management action: Set repurchase gates before execution: minimum liquidity, funded downside investment, valuation discipline, dilution targets, and a cadence that can slow without creating a credibility problem.

What was authorized—and what was not

Reuters reported that the board increased Nvidia’s existing repurchase authorization by $150 billion, taking remaining capacity to $235 billion through the end of fiscal 2028. Reuters said the authorization surpassed Apple’s $110 billion plan from 2024 and exceeds the market value of roughly 84% of S&P 500 companies.

The board did not commit Nvidia to spend $235 billion on a fixed schedule. An authorization creates permission and capacity; it is not a liability, forecast, or promise. Nvidia can buy none, some, or all of the amount, and it can change the pace as cash generation, valuation, and investment needs evolve.

The distinction matters because headline authorization and realized shareholder return are different numbers. Nvidia’s May 20 quarterly release disclosed an earlier $80 billion authorization. Its accompanying CFO commentary said the company returned about $20 billion to shareholders during the quarter through repurchases and dividends. The authorization was much larger than near-term execution even before Monday’s increase.

Record size can obscure the real decision

A board should call cash “excess” only after it funds operating needs, contractual commitments, research and strategic investment, a minimum liquidity reserve, and a credible downside case. That sequencing is especially important during a capital cycle in which customers are financing enormous AI infrastructure programs and suppliers face unusually rapid product and competitive change.

Nvidia can rationally invest heavily and repurchase shares at the same time if cash generation remains strong enough. But the board still has to stress-test timing. Revenue growth can slow before investment commitments do. Margins can compress while strategic spending remains essential. Liquidity that looks idle in the base case can become option value in the downside case.

That is why a large authorization should not be evaluated by dividing it into today’s cash balance or treating every authorized dollar as already surplus. The program spans future periods and can be funded from future cash generation. The opposite shortcut is also dangerous: a large authorization is not proof that future cash will arrive on schedule.

Repurchases can return surplus—or mask dilution

Nvidia reported in its fiscal 2026 annual report that it repurchased 282 million shares for $40.4 billion during that year. The company has also previously described repurchases as a way to offset dilution from employee equity awards while preserving liquidity.

Gross repurchase dollars therefore are not enough to judge the result. Investors and directors should also track the net change in diluted shares, stock-based compensation, average purchase price, and the portion of buybacks that actually reduces the ownership denominator. A program that merely absorbs employee issuance may be economically sensible, but it is different from a program that meaningfully shrinks share count.

Price still matters

Reuters cited LSEG data showing Nvidia at about 16.5 times forward earnings, its lowest forward multiple since January 2015 and below its 15-year average. That is an estimate built on consensus forecasts, not a fixed measure of intrinsic value. It can fall because price declines—or because earnings expectations prove too optimistic.

Repurchases create value when shares are bought below a reasonable estimate of intrinsic value. They destroy value when a company overpays, even if earnings per share rises mechanically. The hurdle is also higher when safe assets yield more: as discussed in When the Risk-Free Rate Hits 5%, Every Capital Allocation Decision Changes, treasury returns raise the opportunity cost of every capital deployment.

Five questions for the board

1. What is truly excess? Define cash remaining after the base plan, committed investment, minimum liquidity, and the downside case.
2. What valuation range triggers action? Set conditions for accelerating, slowing, or suspending repurchases before market volatility forces the debate.
3. How much offsets dilution? Separate gross repurchase spending from net share-count reduction.
4. What cash-flow evidence supports the horizon? Compare the authorization with base and downside free-cash-flow scenarios through fiscal 2028.
5. What would make management stop? A flexible authorization is valuable only if the company is willing to preserve flexibility.

This is the same discipline required for technology investment. AI ROI Is a Finance Problem, Not a Technology Problem argues that cash flows, counterfactuals, and opportunity cost should control the decision. The framework applies whether capital is going into a new system or back to shareholders. The free Capital Project ROI & Payback tool can help make those alternatives explicit.

The bottom line

The headline is $150 billion. The judgment is whether each dollar remains surplus at the moment it is spent. A disciplined board will treat the authorization as an option whose exercise depends on liquidity, strategic investment, dilution, valuation, and new evidence—not as a target that must be met.

Sources

• Reuters, September 28, 2026
• Associated Press, September 28, 2026
• Nvidia fiscal 2027 first-quarter release, May 20, 2026
• Nvidia fiscal 2026 annual report

Reported facts and estimates are attributed above. The judgments about excess cash, valuation discipline, dilution, and board governance are Numbers & Judgment analysis.