breaking February 25, 2026

Nvidia Shatters Records Again: $68B Quarter Confirms AI Infrastructure Boom Has No Ceiling

Nvidia’s fiscal Q4 2026 results posted record revenue of $68.1 billion (up 73% year-over-year) as data center demand from AI hyperscalers obliterated analyst estimates and Q1 guidance of $78B silenced AI bubble fears.

Nvidia delivered another seismic earnings report Wednesday, posting fiscal fourth-quarter revenue of $68.13 billion, a 73% jump from the same period a year ago, obliterating Wall Street estimates and erasing any remaining doubt about the durability of the AI infrastructure buildout. Markets rallied Thursday as investors digested numbers that turned AI capex fatigue fears into noise.

The Numbers That Rewrote the Record Books

The headline figures were staggering even by Nvidia’s own recent standards. Adjusted earnings per share came in at $1.62, beating the $1.53 consensus. Revenue beat estimates of $66.21 billion by nearly $2 billion. Net income almost doubled year-over-year to $43 billion.

The most significant signal was inside the data center division, which now accounts for over 91% of Nvidia’s total revenue. Data center revenue reached $62.3 billion for the quarter, a 75% year-over-year increase, beating the Street’s $60.69 billion forecast. The concentration of the business in AI compute is now so complete that Nvidia is effectively a pure-play AI infrastructure company wearing a GPU brand.

Networking was the hidden standout. Revenue from Nvidia’s networking products, including NVLink and Spectrum-X Ethernet switches used to connect massive GPU arrays, surged 263% year-over-year to $10.98 billion. That growth reflects the scale at which hyperscalers are building rack-scale AI clusters rather than standalone servers.

Jensen Huang Signals There Is No Plateau

CEO Jensen Huang’s message to investors was characteristically unambiguous: the demand curve is not bending. He noted that Nvidia’s data center business has grown 13x since ChatGPT launched in late 2022, and underscored that hyperscalers, including Alphabet, Amazon, Meta, and Microsoft, are not slowing their commitment. Combined capital expenditure projections for the four major cloud providers this calendar year could approach $700 billion as they race to build AI infrastructure at unprecedented scale.

The Q1 fiscal 2027 guidance of $78 billion, against analyst expectations of $72.6 billion, functioned as a hard rebuttal to the “AI capex fatigue” narrative that had briefly rattled markets in January. Pre-earnings, the VIX had spiked nearly 18%, breaching the 20 threshold as investors worried hyperscalers might finally blink on AI spending. The guidance made clear they have not.

Huang also confirmed on Wednesday’s call that Nvidia shipped its first Vera Rubin samples to customers earlier this week. Vera Rubin is the next-generation rack-scale architecture succeeding Grace Blackwell, with production shipments scheduled to begin in the second half of 2026. Excitement around Vera Rubin has been building for months; the platform is expected to redefine the performance ceiling for large-scale AI training once it reaches volume.

Constraints and the Road Ahead

Not everything ran at full throttle. Nvidia’s gaming unit, once its largest segment, recorded a 13% quarter-over-quarter revenue decline to $3.7 billion, even as it grew 47% from a year ago. Analysts have speculated that Nvidia may skip a dedicated new gaming GPU launch this year as a global shortage of high-bandwidth memory forces the company to prioritize AI accelerators. CFO Colette Kress confirmed in her written commentary that supply constraints are expected to continue pressuring the gaming business into the first quarter of FY2027 and beyond.

The memory shortage is a structural watch item for the industry. With SK Hynix’s stock having nearly quadrupled over six months while Nvidia’s own shares gained a comparatively modest 5% year-to-date, the market is beginning to price supply chain leverage into the AI infrastructure value chain, a dynamic that may intensify as Vera Rubin demand materializes.

Why This Quarter Matters Beyond the Numbers

Nvidia’s results function as a real-time proxy for the entire AI infrastructure stack. When Nvidia beats at this scale, with data center sales up 75% and guidance that clears consensus by more than $5 billion, it confirms that every layer of the AI buildout, from chips to interconnects to cooling to energy, remains in growth mode. For AI companies, model labs, and enterprise buyers, the message is consistent: the foundation being poured right now is significantly larger than the previous cycle’s peak, and it is not finished.

See also: AI Copyright Cases Reach the Supreme Court | X01.

For related context, see NVIDIA Rubin and N1X: Rewriting the Rules of AI Hardware.

The Q1 fiscal 2027 guidance of $78 billion, against analyst expectations of $72.6 billion, functioned as a hard rebuttal to the “AI capex fatigue” narrative that had briefly rattled markets in January. Pre-earnings, the VIX had spiked nearly 18%, breaching the 20 threshold as investors worried hyperscalers might finally blink on AI spending. The guidance made clear they have not.

Huang also confirmed on Wednesday’s call that Nvidia shipped its first Vera Rubin samples to customers earlier this week. Vera Rubin is the next-generation rack-scale architecture succeeding Grace Blackwell, with production shipments scheduled to begin in the second half of 2026. Excitement around Vera Rubin has been building for months; the platform is expected to redefine the performance ceiling for large-scale AI training once it reaches volume.

Constraints and the Road Ahead

Not everything ran at full throttle. Nvidia’s gaming unit, once its largest segment, recorded a 13% quarter-over-quarter revenue decline to $3.7 billion, even as it grew 47% from a year ago. Analysts have speculated that Nvidia may skip a dedicated new gaming GPU launch this year as a global shortage of high-bandwidth memory forces the company to prioritize AI accelerators. CFO Colette Kress confirmed in her written commentary that supply constraints are expected to continue pressuring the gaming business into the first quarter of FY2027 and beyond.

The memory shortage is a structural watch item for the industry. With SK Hynix’s stock having nearly quadrupled over six months while Nvidia’s own shares gained a comparatively modest 5% year-to-date, the market is beginning to price supply chain leverage into the AI infrastructure value chain, a dynamic that may intensify as Vera Rubin demand materializes.

Why This Quarter Matters Beyond the Numbers

Nvidia’s results function as a real-time proxy for the entire AI infrastructure stack. When Nvidia beats at this scale, with data center sales up 75% and guidance that clears consensus by more than $5 billion, it confirms that every layer of the AI buildout, from chips to interconnects to cooling to energy, remains in growth mode. For AI companies, model labs, and enterprise buyers, the message is consistent: the foundation being poured right now is significantly larger than the previous cycle’s peak, and it is not finished.

The AI capex super-cycle has passed its first serious test. It passed decisively.