Nvidia, the world's most valuable company with a market capitalization of approximately 6,805.836 trillion won, reported quarterly results on February 25 that once again exceeded market expectations, silencing concerns about an artificial intelligence bubble and raising the bar for future performance.
"Computing demand is growing exponentially," CEO Jensen Huang said, signaling even greater growth ahead. He attributed this to surging demand for computing semiconductors that support reasoning capabilities—the core function of agentic AI assistants.
The chipmaker posted revenue of $68.127 billion for the fourth quarter of fiscal 2026 (November 2024 to January 2025), up 73% year-over-year and surpassing the consensus estimate of $66.2 billion. Annual revenue reached $215.9 billion, a 65% increase from the prior year.
Data center chip sales drove the results, jumping 75% to $62.314 billion and accounting for roughly 90% of total revenue. Net profit margin hit a record 63% for the period.
Huang predicted the proliferation of agentic AI will trigger explosive growth in computing and chip demand. Unlike traditional large language models, agentic AI systems plan and execute tasks autonomously, repeatedly running inference processes. Industry estimates suggest these systems require three to ten times more AI chips than conventional LLMs.
"The Grace Blackwell AI accelerator is currently the leader in inference," Huang said. "Vera Rubin will further expand our dominance."
For the current quarter (February to April), Nvidia projects revenue of $78 billion, well above the Wall Street consensus of $72.6 billion.
Cash reserves have also swelled. The company holds approximately $62.6 billion (about 89.399 trillion won) in cash and equivalents, up 45% from $43.2 billion a year earlier.
Huang pushed back against criticism that AI infrastructure investment by customers including Alphabet, Amazon, and Meta has become excessive. "In the new AI world, computing equals revenue, so their cash flows will grow," he said. "Considering all their current resources, I'm confident they hold roughly $300 billion to $400 billion in cash."
Market concerns persist that massive customer investments may hit limits after next year, potentially slowing Nvidia's growth. Shares fluctuated, falling as much as 1.5% before closing barely positive at 0.1%.
Still, some analysts are optimistic about Nvidia's cash-generation capacity. QuickFactSet projects the company will generate $168 billion in free cash flow this year. Nikkei noted this exceeds several Big Tech companies combined.
Memory chip shortages pose a potential obstacle. Nvidia's AI chips require memory semiconductors, and recent supply constraints have driven prices sharply higher, contributing to "chipflation" that raises costs across products. Bloomberg called the memory shortage "one dark cloud that could engulf Nvidia."
Nvidia said it has secured "strategic inventory and production capacity to meet demand for several quarters ahead." However, the company acknowledged its gaming division could face difficulties after the first quarter of 2027 due to slower production.
Critics have also flagged "circular transactions" in which Nvidia invests in tech companies that then purchase its chips. Steve Hanke, professor at Johns Hopkins University, said: "Part of Nvidia's revenue is being utilized this way—this is not naturally occurring demand." Nvidia recently reduced such arrangements with OpenAI from $100 billion to $30 billion. Nikkei expressed concern that tech companies receiving Nvidia investments may struggle to recoup their massive AI spending.
Huang also addressed the disputed potential of space-based data centers. "The economics aren't favorable now, but will improve over time," he said. "The best use case for GPUs in space is imaging—capturing extremely high-resolution images and transmitting petabyte-scale data to Earth is far simpler from space than from the ground."