• NVIDIA (NVDA)'s board authorized an additional $80 billion for share repurchases in May 2026, not the $150 billion increase implied by recent headlines. The total authorization stands at $235 billion, but roughly $99 billion remained available after fiscal Q2 2027 capital returns.
  • The company has not stated that the buyback must be completed by fiscal 2028; the May authorization carries no expiration. NVIDIA's aggressive capital return comes alongside extraordinary AI-driven cash generation, with fiscal Q2 revenue hitting $96.2 billion, up 106% year over year.
  • Management projected fiscal Q3 revenue of about $108 billion and forecast approximately 70% revenue growth in fiscal 2028, well above Wall Street's 44% expectation, even as memory shortages and higher component costs pressure gross margins.

NVIDIA's Buyback Math: What's Real and What's Not

NVIDIA Corp. is returning capital to shareholders at a pace that rivals its breakneck revenue growth, but recent headlines touting a $235 billion buyback expansion have conflated the company's total authorization with a single board action. The actual confirmed development: In May 2026, NVIDIA's board added $80 billion to its repurchase program and raised the quarterly dividend from $0.01 to $0.25 per share. That authorization has no expiration date, according to company disclosures. Reports suggesting a $150 billion increase and a fiscal 2028 completion deadline do not match the primary filings reviewed as of September 28, 2026.

The distinction matters for investors parsing NVIDIA's capital allocation strategy. A buyback authorization is a ceiling, not a commitment. After returning roughly $26 billion through repurchases and dividends in fiscal Q2 2027 alone, NVIDIA had approximately $99 billion remaining under its existing authorizations. That's a substantial war chest, but it reflects the company's rapid execution rather than a single headline-grabbing figure.

AI Cash Machine Funds Both Growth and Returns

NVIDIA's ability to fund buybacks while investing heavily in next-generation platforms stems from its unprecedented cash generation. In fiscal Q2 2027, ended July 26, revenue reached $96.2 billion, up 106% year over year and 18% sequentially, with Data Center revenue at $89.0 billion, up 117%. Fiscal Q1, ended April 26, saw revenue of $81.6 billion, up 85%, and Data Center revenue of $75.2 billion, up 92%.

The company is balancing three priorities: supplying rapidly growing data-center demand, investing in platforms such as Vera Rubin, and returning capital to shareholders. Management projected fiscal Q3 revenue of about $108 billion, plus or minus 2%, versus a $104.19 billion analyst consensus, and unusually forecast approximately 70% revenue growth in fiscal 2028, well above the 44% Wall Street expectation reported by Reuters. The Vera Rubin platform has begun shipping and is expected to contribute about one-fifth of Data Center revenue in the quarter ending October.

Margin Pressure and Supply Constraints Loom

Even as demand surges, NVIDIA faces near-term profitability headwinds. Management expects shortages and higher prices for memory and other components to compress gross margin, indicating margins could bottom around 71%–72% in fiscal Q4, versus roughly 74% in fiscal Q3. The company says it remains supply-constrained, a dynamic that can delay customer deployments and raise computing costs across the AI ecosystem.

NVIDIA and Amazon Web Services (AMZN) announced plans to deploy another 2 million NVIDIA GPUs across Amazon's global infrastructure in 2027 and 2028. Reuters reported that AI labs are expected to account for about one-quarter of NVIDIA's overall business next year, while demand broadens beyond hyperscale-cloud customers to enterprises, sovereign buyers, industrial customers, and specialist AI clouds. That broadening supports the bullish case, but it also increases reliance on a small group of high-capacity suppliers and advanced packaging/memory availability.

China Remains a Wild Card

NVIDIA's China exposure remains strategically significant but highly uncertain. The U.S. formally permitted certain H200 exports to China in January 2026 under conditions including third-party technical review and restrictions on military use. U.S. Commerce later clarified that licensing obligations also apply to China-headquartered entities operating outside China. Beijing, meanwhile, has constrained demand: reports indicated authorities limited H200 purchases, and by August only small shipments were reaching mainland China. Reuters reported that ByteDance and Tencent (TCEHY) had received roughly 10,000 H200 processors each, while Beijing preferred that some hardware remain outside the mainland to encourage domestic semiconductor development. NVIDIA excludes China data-center revenue from its outlook, making it a material upside/downside variable rather than a core assumption.

The Bottom Line

NVIDIA's capital return program has expanded alongside its AI earnings: an additional $50 billion in buyback authority in August 2024, another $80 billion in May 2026, and a 25-fold dividend increase. The confirmed recent development is the $80 billion May addition with no stated expiration, leaving roughly $99 billion available after fiscal Q2. Investors should treat the '$235 billion through fiscal 2028' framing as unverified. The real story is a company returning enormous sums while financing an exceptionally large AI-infrastructure growth cycle—and facing real risks from memory supply, margin compression, and geopolitical constraints. NVIDIA did not respond to a request for comment on the buyback figures.

Correction: An earlier version of this article misstated the size of the May 2026 buyback authorization. It was $80 billion, not $150 billion.