Nvidia (NVDA) shares climbed roughly 3% on Friday, touching a fresh intraday record and pulling ahead of the chipmaker's previous all-time closing high from May. Shares traded near $237.75 during the session, extending a rally that has lifted the stock about 27% this year.

The advance drew strength from two directions at once. One is a brewing argument over how long AI chips actually last, a dispute that goes to the heart of how Big Tech's massive AI investments are accounted for. The other is cold, hard cash heading back to shareholders. Both themes converged in Friday's rally.

Nvidia Counters the Depreciation Skeptics

At the center of the dispute is investor Michael Burry, best known for calling the 2008 housing crash, who has argued that major technology companies are underestimating how quickly their AI chips wear out. If he is right, today's profits would be propped up by accounting that flatters the numbers.

The accounting mechanics are what make the argument consequential. Companies spread the cost of servers across the number of years they expect them to stay in service, so a longer assumed lifespan means less expense charged against earnings each year — and a larger reported profit from the same spending.

Nvidia is not staying quiet on the issue. The company is highlighting that major customers such as Meta Platforms and Google have stretched their AI server depreciation schedules from four years to six years. That is a direct rebuttal aimed at the core of the skeptics' case, and it signals that Nvidia intends to contest the criticism rather than let it go unanswered.

Why the $150 Billion Buyback Matters

The second driver is more straightforward. Nvidia's board authorized an additional $150 billion in share repurchases this week, bringing the company's total remaining buyback authorization to $235 billion through fiscal 2028. That is the largest buyback authorization in corporate history.

Buybacks return cash to shareholders by repurchasing stock, and a standing authorization of this size gives the company room to keep doing so over multiple years.

The scale of the program is no small statement of confidence. It signals that Nvidia's management believes the company can keep generating serious cash even while pouring money into AI infrastructure.

Wall Street took notice. Morgan Stanley reinstated Nvidia as its top semiconductor pick on Friday following meetings with CEO Jensen Huang and other executives, pointing to continued strength in AI demand and a growing customer base.

Morgan Stanley also flagged something worth watching. The firm argued that the AI infrastructure bottleneck is shifting away from chip production and toward data center land, power supply, and financing. That marks a meaningful change in the conversation: if chips are no longer the scarce resource, the companies holding onto cheap power and prime real estate stand to gain a bigger edge.

The Bigger AI Picture

Nvidia is not operating in a vacuum. The broader AI trade has investors looking past pure infrastructure spending toward what comes next in the technology cycle. Anthropic's expected initial public offering is adding fuel to that fire: a public listing for the developer of the Claude chatbot would serve as another high-profile test of how much appetite investors still have for AI companies and the capital they need to scale up.

For Nvidia specifically, the combination of a record buyback and a public counterargument to its critics gives the stock a fresh story heading into October. The company is no longer simply letting the depreciation debate play out in the background.

How the dispute unfolds — and whether more of Nvidia's largest customers follow Meta and Google in extending their schedules — now sits at the center of how investors read Big Tech's AI-era earnings.

Nvidia also works closely with Wall Street firms to help deploy roughly $500 billion in capital supporting the AI sector, a figure that underscores just how large the stakes have become in this debate over chip lifespans and spending.