Nvidia $150 Billion Buyback Sets a Record as AI Doubt Grows

Nvidia authorized a record $150 billion buyback increase, lifting its remaining repurchase capacity to $235 billion as doubts about the AI investment boom deepen.

The announcement landed during a volatile US trading session on September 28. Oil jumped after President Donald Trump rejected Iran’s proposal for reopening the Strait of Hormuz. That geopolitical shock dominated the morning, but Nvidia’s capital decision delivered the day’s clearest technology signal.

The Nvidia $150 billion buyback is not simply a reward for shareholders. It is a wager that Nvidia can finance new chips, supply commitments, strategic investments, and an unprecedented capital-return program at the same time. That promise now faces a harder test from AMD, Intel, custom silicon, and increasingly cautious AI infrastructure investors.

What Nvidia’s Record Authorization Actually Changes

Nvidia did not spend $150 billion in one transaction. Its board expanded the amount available for future repurchases.

The board added $150 billion to Nvidia’s existing share-repurchase authorization on September 28. That brought the total remaining authorization to $235 billion, according to the company’s buyback announcement.

Nvidia expects to execute the remaining program through its 2028 fiscal year, which ends in January 2028. The authorization gives management permission to repurchase shares, but it does not create a fixed purchase schedule.

That distinction matters. Nvidia can adjust the pace according to its share price, cash position, operating needs, and other investment opportunities. It can buy shares in the open market or through structured plans, subject to applicable rules.

The increase is still exceptional. Nvidia called it the largest share-repurchase authorization increase in history. It surpassed Apple’s $110 billion authorization from 2024, which had set the previous corporate record.

The new approval arrived only four months after Nvidia’s board authorized another $80 billion. At the end of July, the company still had $99.3 billion available under its previous program.

Nvidia was already moving quickly. Its latest quarterly filing shows that it repurchased 94 million shares for $19.7 billion during the second quarter. It bought 203 million shares for $39.8 billion across the first half.

Those figures explain why another authorization arrived so soon. Nvidia had the capacity to keep purchasing shares, but its current pace could have consumed a substantial part of that capacity before fiscal 2028.

The company returned about $26 billion to shareholders through repurchases and dividends during its latest quarter. Repurchases accounted for most of that amount.

CEO Jensen Huang framed the new authorization as evidence that Nvidia can invest and return capital simultaneously. He said the company’s cash generation supports technology investment while allowing it to reward shareholders.

That statement defines the central promise behind the Nvidia $150 billion buyback. Nvidia is telling investors that capital returns do not compete with product development, manufacturing commitments, or ecosystem investments.

Markets gave the announcement an initially positive response. Nvidia shares rose more than 2% during Monday trading, according to market reporting.

However, the share move was modest beside the authorization’s scale. Investors appeared to recognize both sides of the decision. Nvidia possesses an unusual cash engine, but it is deploying that cash during an unusually uncertain phase of the AI cycle.

The announcement therefore changed more than Nvidia’s capital-return ceiling. It converted management’s confidence in sustained AI demand into a measurable commitment with a deadline.

Why Nvidia Can Make a $235 Billion Commitment

The authorization rests on Nvidia’s current earnings power, not on financial engineering alone.

Nvidia reported $96.2 billion in revenue for the quarter ending July 26, up 106% from the previous year. Revenue increased 18% from the preceding quarter.

Data Center revenue reached $89 billion, a 117% annual increase. That division supplied more than nine-tenths of Nvidia’s total quarterly revenue.

The company produced $59.7 billion in GAAP net income. Its operating income reached $63.7 billion, while its gross margin stood at 75%.

These numbers place the buyback in context. The $150 billion increase equals about two and a half times Nvidia’s latest quarterly net income. It is enormous by corporate standards, yet it is not detached from the company’s recent profitability.

Nvidia’s quarterly results also reported $21.3 billion in free cash flow. Free cash flow measures cash generated after capital spending and offers a practical view of what remains available for other uses.

That quarterly cash figure does not cover a $235 billion authorization by itself. Nvidia is instead betting on repeated cash generation through fiscal 2028.

Its near-term outlook supports that case. Nvidia forecast $108 billion in third-quarter revenue, with a permitted variation of 2%. The forecast excluded any Data Center compute revenue from China.

Management has also forecast about 70% revenue growth for fiscal 2028. That unusual long-range projection gave investors a concrete basis for assessing the buyback’s execution window.

The company ended its July quarter with $22.4 billion in cash and cash equivalents. It also held substantial marketable securities and strategic equity investments.

Nvidia’s balance sheet, however, is more complex than a simple cash total suggests. Some investments face lockup restrictions, while other holdings can fluctuate with public and private market valuations.

The company also carries major operating commitments. Nvidia depends on third parties to manufacture, package, and test its processors. Securing advanced memory, fabrication capacity, networking components, and complete rack systems requires large forward commitments.

Inventories reached $31.6 billion at the end of July, up from $21.4 billion in January. That increase reflects the physical scale of Nvidia’s expanding product pipeline.

The company reported $47.9 billion in nonmarketable equity securities. It also held $3.3 billion in infrastructure-financing investments accounted for under the equity method.

Those investments are part of Nvidia’s broader effort to accelerate AI infrastructure construction. They can expand the market for its processors, but they also tie Nvidia more closely to its customers’ financing conditions.

Nvidia disclosed that five direct customers represented 22%, 14%, 13%, 11%, and 10% of its accounts-receivable balance. That concentration creates another reason to scrutinize future cash conversion.

The company sometimes provides investment-grade customers with payment periods ranging from 90 days to one year for large data center projects. Longer terms can support deployment, although they also delay cash collection.

None of these commitments invalidates Nvidia’s capital-return decision. They show why cash flow, rather than reported net income alone, should guide analysis of the program.

The Nvidia $150 billion buyback signals confidence because management expects the company’s cash engine to outrun these demands. Investors must now test that expectation against each reporting period.

The Nvidia $150 Billion Buyback Faces an AI Demand Test

The record authorization is most meaningful because Nvidia’s valuation shows that investors still doubt the durability of its growth.

At the time of the announcement, Nvidia traded at about 16.5 times expected earnings for the following 12 months. LSEG data cited by Reuters placed that multiple at its lowest level since January 2015.

The same data put Nvidia’s 15-year average near 30 times forward earnings. Its current multiple therefore reflected substantial skepticism, despite the company’s extraordinary revenue and profit growth.

That combination is the central reversal in this story. Nvidia is generating more money than ever, yet investors assign a lower earnings multiple because they expect growth to slow.

A buyback can be particularly effective when management considers a company’s shares undervalued. Repurchasing stock at a lower valuation retires more shares for the same amount of cash.

Fewer shares can increase earnings per share, assuming earnings remain constant. That mathematical benefit does not create new operating profit, but it can increase each remaining shareholder’s claim on existing profit.

The board’s decision suggests that Nvidia sees attractive value in its stock. However, an authorization is not a precise declaration that management believes the current market price is wrong.

Nvidia can vary the repurchase pace, and it has not disclosed an average target price. Investors should avoid treating the full $235 billion as an immediate bid beneath the stock.

The valuation discount reflects a broader argument about AI infrastructure. Cloud providers, AI laboratories, governments, and enterprises have spent heavily on processors and data centers. The market now wants evidence that this spending can produce durable returns.

Nvidia sits at the profitable center of that buildout. Its customers bear much of the cost, while Nvidia sells chips, networking equipment, systems, and software needed to run AI workloads.

That position produced exceptional margins during the construction phase. The next question is whether AI services generate enough revenue to sustain the purchasing cycle.

Huang argues that compute has become a direct source of revenue. Nvidia says multiple frontier laboratories, startups, open-model developers, and physical AI companies are now driving demand.

The claim is plausible, but the company’s quarterly results cannot establish the ultimate economics of its customers. Strong chip sales measure infrastructure demand, not necessarily the long-term profitability of every AI service using those chips.

Industry spending has also become more dependent on debt, vendor financing, and long-term capacity agreements. The structure can extend the buildout, while increasing sensitivity to interest rates and customer cash flow.

That is why the buyback represents more than excess cash distribution. Nvidia is committing to return capital while its customers continue financing projects that support Nvidia’s growth.

The authorization becomes a stronger confidence signal if Nvidia can sustain repurchases without weakening research, supply access, or strategic flexibility. It becomes less persuasive if customer financing deteriorates or free cash flow falls far behind earnings.

The debate therefore turns on execution. Nvidia’s record does not settle the AI demand question. It raises the financial stakes around management’s answer.

AMD, Intel, and Custom Chips Raise the Pressure

Nvidia remains the AI chip leader, but its buyback arrives as competitors gain performance, customers, and investor attention.

Nvidia shares had gained just over 20% during 2026 through September 25. That roughly matched the Nasdaq 100, but it trailed several semiconductor competitors.

AMD shares had more than doubled during the same period. Intel shares had more than tripled, according to Reuters.

Stock performance does not measure market share directly. It does show where investors expect the largest change in future earnings.

Nvidia begins from a much stronger revenue and profit position than either rival. Its CUDA software platform, high-speed networking, rack-scale systems, and developer ecosystem create switching costs beyond processor specifications.

Customers do not choose only a chip. They choose software libraries, interconnects, orchestration tools, support, and an upgrade path for costly infrastructure.

AMD is attacking that structure with its Instinct accelerators and an expanding software stack. Large customers have a financial incentive to qualify alternative suppliers, even when Nvidia remains their primary platform.

A viable second source can improve customers’ negotiating leverage and reduce their exposure to shortages. It can also place pressure on Nvidia’s margins if comparable workloads become portable across competing hardware.

Intel’s challenge comes from several directions. Its accelerators compete for selected AI workloads, while its foundry ambitions seek a position in the manufacturing chain. Its recent stock performance reflects expectations of improvement rather than parity with Nvidia’s AI business.

The hyperscale cloud companies create another competitive route. Amazon, Google, and Microsoft design custom accelerators that target workloads inside their own data centers.

Custom silicon does not need to replace Nvidia everywhere to affect the market. It can capture stable, high-volume workloads where cloud operators control the software and can optimize around one architecture.

AI laboratories are also investing in software efficiency. Better model architectures, lower numerical precision, and improved scheduling can reduce the amount of compute required for a given task.

Efficiency can expand demand by making AI cheaper, but it can also slow hardware growth if workloads do not expand quickly enough. Nvidia’s thesis assumes that lower costs unlock much greater usage.

The company is responding by broadening its platform. It sells complete systems rather than isolated graphics processors, while extending into networking, CPUs, inference products, robotics, and enterprise software.

This strategy helps Nvidia capture more value from each deployment. It also increases the number of markets where the company must execute successfully.

The buyback introduces a visible comparison between cash returned to shareholders and cash directed toward expansion. Investors will ask whether Nvidia is purchasing stock because its opportunities remain abundant, or because fewer internal projects meet its return requirements.

There is no evidence that Nvidia has stopped investing. Its operating expenses reached $8.4 billion in the latest quarter, up 55% from a year earlier.

The company also continues to commit funds to suppliers, customers, and infrastructure partners. The relevant issue is not whether Nvidia invests, but whether its capital allocation remains optimal as competition widens.

Apple offers the closest historical comparison. It paired enormous repurchases with a mature, highly profitable hardware and services business. Analyst Ben Barringer said Nvidia’s decision echoes that playbook as growth and valuation multiples face pressure.

The resemblance has limits. Apple’s cash flows come from a large consumer ecosystem with recurring replacement cycles. Nvidia’s recent growth depends more heavily on concentrated infrastructure spending.

That difference makes Nvidia’s record authorization more aggressive. The company is adopting mature-company capital returns while operating inside a young and volatile market.

What the Buyback Numbers Do Not Prove

A large authorization proves that Nvidia has financial confidence, but it does not prove that AI demand or Nvidia’s market position will remain unchanged.

The first uncertainty is execution. Nvidia expects to use the authorization through fiscal 2028, but it has not promised equal purchases each quarter.

A falling share price might cause Nvidia to accelerate purchases. Operating needs, acquisitions, supply commitments, or market conditions could cause it to slow them.

The second uncertainty is dilution. Share repurchases reduce the share count only when purchases exceed new shares issued through employee compensation and other programs.

Nvidia’s share count has recently declined, indicating that repurchases have exceeded dilution. Investors should continue checking that outcome rather than focusing only on gross spending.

The third uncertainty concerns cash conversion. Nvidia reported $59.7 billion in quarterly net income but generated $21.3 billion in free cash flow.

Several timing factors can produce a gap between those figures. Working capital, taxes, investment settlements, and other movements can shift cash generation between reporting periods.

One quarter does not establish a trend. A sustained gap, however, would make the repurchase plan more dependent on balance-sheet assets or future cash improvement.

The fourth uncertainty involves customer concentration. Nvidia sells into a relatively small group of large buyers with the resources to construct advanced data centers.

Those customers also have the greatest motivation to reduce dependence on one supplier. They can negotiate pricing, build custom chips, optimize models, or delay projects when returns disappoint.

The fifth uncertainty is geopolitical access. Nvidia excluded China Data Center compute revenue from its third-quarter forecast. Export restrictions can close important markets or require the company to redesign products.

The Strait of Hormuz crisis added a separate macroeconomic risk when the buyback was announced. Oil climbed after Trump rejected Iran’s reopening proposal, which called for concessions involving sanctions, frozen funds, and military pressure.

The Hormuz standoff matters to Nvidia because energy prices affect data center operating costs, inflation expectations, interest rates, and technology valuations.

An energy shock would not erase demand for AI processors. It could make financing more expensive for the companies building the infrastructure around them.

The sixth uncertainty is opportunity cost. Every dollar used for a repurchase becomes unavailable for another purpose at that moment.

Nvidia must balance shareholder returns against research, acquisitions, supply agreements, manufacturing access, and ecosystem support. A good buyback requires more than sufficient cash. It requires the stock purchase to offer an attractive return against those alternatives.

Management’s incentives deserve attention here. Higher earnings per share can support valuation and compensation metrics, even when repurchases do not improve the underlying business.

That does not make the program misguided. It means investors should judge its effectiveness through per-share results, not through the authorization’s headline size.

The strongest version of Nvidia’s argument is straightforward. Its platform produces enough cash to fund leading research, secure supply, expand the market, and retire shares at an attractive valuation.

The skeptical version is equally clear. Nvidia is supporting its stock as the market prices in slower growth and stronger competition.

Both readings can contain truth. Capital allocation often serves several goals at once.

The decisive evidence will come from Nvidia’s future cash flow, share count, margins, and customer demand. The announcement alone cannot resolve those variables.

Three Signals Will Show Whether Nvidia’s Bet Works

The next several months will reveal whether the record authorization reflects durable cash generation or confidence near the top of an investment cycle.

The first signal is Nvidia’s actual repurchase pace.

The company spent $19.7 billion buying shares during its latest quarter. Investors should compare the next two quarters with that baseline and track the average price paid.

Accelerating purchases while maintaining strong free cash flow would reinforce management’s valuation argument. A sharp slowdown would suggest that operating commitments or market conditions require greater caution.

The net share count matters more than gross spending. Repurchases that merely offset employee compensation provide less benefit than purchases that produce sustained per-share contraction.

Investors should therefore watch diluted shares outstanding alongside the cash-flow statement. That combination reveals whether the program is increasing each shareholder’s economic claim.

The second signal is Data Center growth and cash conversion.

Nvidia’s $89 billion in quarterly Data Center revenue provides the foundation for the buyback. Revenue must continue converting into operating and free cash flow as inventories and supplier commitments expand.

Third-quarter revenue near Nvidia’s $108 billion forecast would support the near-term demand case. The more important test will be whether customers maintain orders into fiscal 2028.

Cloud-company capital spending offers an early indicator. Stable or rising budgets would strengthen Nvidia’s forecast, while project delays or lower spending plans would weaken it.

Receivables also deserve attention. Nvidia’s longer payment terms can help customers complete major builds, but a rapid expansion would raise questions about the quality and timing of cash collection.

The third signal is competitive adoption outside Nvidia’s platform.

AMD accelerator deployments, Intel customer wins, and new cloud-designed processors will show whether customers are creating credible alternatives. The relevant measure is sustained production use, not demonstration benchmarks.

A competitor does not need to overtake Nvidia across the market. It only needs to capture enough high-volume workloads to affect pricing, growth expectations, or customer bargaining power.

Nvidia can answer that pressure through better performance, lower total operating costs, stronger software, and faster product cycles. Its integrated systems may also preserve an advantage when deployment speed matters more than individual chip prices.

The Nvidia $150 billion buyback will look well judged if those three signals move together. Repurchases must reduce the share count, cash flow must support them, and Nvidia’s platform must retain its economic advantage.

If only the share purchases remain strong, the authorization risks becoming a defense against weaker fundamentals. If operating results remain strong while repurchases slow, management may have found higher-return uses for the cash.

That distinction is why readers should resist treating the record amount as a final verdict. The announcement is the beginning of a capital-allocation test that runs through January 2028.

Nvidia has converted confidence into a public commitment. Now investors, developers, and enterprise technology buyers should follow the evidence behind it.

Watch each earnings report for cash flow, customer concentration, Data Center growth, and the net share count. Those figures will show whether Nvidia is buying an undervalued cash machine or supporting a valuation before the AI cycle changes.