Nvidia posts $68.1B Q4 revenue; raises AI outlook

Record $68.1B driven by data center AI demand and inference

NVIDIA Corporation (NASDAQ: NVDA) reported record revenue of $68.1 billion for fiscal Q4 2026 ended January 25, 2026, up about 20% from the prior quarter, according to NVIDIA’s official newsroom. The quarter set a new high for the company and underscored momentum in accelerated computing tied to AI-driven workloads.

The performance was led by data center demand, with AI inference workloads emerging as a central driver of deployed capacity and utilization. As reported by The Wall Street Journal, CEO Jensen Huang framed the period as a structural shift in computing tied to agentic AI that sharply expanded profit contribution.

Guidance: about $78B next quarter; supply commitments and visibility rising

For the current quarter, management guided to about $78 billion in revenue and highlighted stronger supply arrangements to meet intensifying customer requirements, as reported by Fortune. The company also pointed to rising visibility into data center buildouts centered on its Blackwell platform and the follow-on Rubin architecture.

Management emphasized that these commitments are intended to secure capacity beyond the near term. “Supply commitments jumped from roughly $50 billion to about $95 billion within a quarter, positioning us to meet demand beyond the next several quarters,” said Colette Kress, CFO at NVIDIA.

Institutional reaction remained constructive. Morgan Stanley said the revenue outlook helps de-risk concerns about a slowdown in AI infrastructure spending, while Barclays, as summarized by TickerDaily, cautioned that sustaining gross margins in the mid-70s could become more challenging as competition increases.

Agentic AI, Blackwell, and Rubin extend NVIDIA’s inference leadership

Agentic AI refers to systems that can plan and execute multi-step tasks autonomously, expanding beyond prompt-response generation to persistent workflows and tools. According to Axios’s coverage of executive commentary, NVIDIA positions its stack around this shift, with the current Blackwell architecture and upcoming Vera Rubin platform designed to extend the company’s already strong position in inference at scale. Inference differs from training in that it emphasizes low-latency, cost-efficient model execution in production, where hardware-software optimization and networking throughput significantly influence total cost of ownership.

“The agentic AI inflection point has arrived,” said Jensen Huang, CEO of NVIDIA. The comment frames a pivot from solely building ever-larger models toward deploying agents that interact with data, tools, and users, driving sustained demand for high-performance inference infrastructure.

As reported by CRN, NVIDIA also highlighted that demand for GPUs and associated systems continues to ramp and described itself as the world’s largest networking business, tying performance leadership to end-to-end platform breadth. That breadth spans compute, interconnects, and software, which together influence utilization rates and economics for enterprise and cloud deployments.

At the time of this writing, NVDA traded in the mid-$190s with a multi-trillion-dollar market capitalization, offered here solely as contextual market background. This market snapshot is not a forecast and should not be interpreted as investment advice.