Nvidia Q4 Surged to $43 Billion in Earnings, Huang Renxun Says AI Agent Inflection Point Has Arrived

21st Century Business Herald Reporter Ni Yuqing

Amid ongoing concerns about an AI bubble, AI giant NVIDIA continues to grow steadily beyond expectations.

On February 26, NVIDIA announced its fiscal year 2026 Q4 financial results (for the period ending January 25, 2026, equivalent to Q4 of the calendar year 2025), with revenue and profit both rising double digits to hit new records.

In Q4, revenue reached $68.127 billion, up 73% year-over-year and 20% quarter-over-quarter; net profit was $42.96 billion, up 94% year-over-year and 35% quarter-over-quarter; gross margin increased to 75%.

(NVIDIA Q4 revenue by business line, Source: NVIDIA official website)

For the full fiscal year 2026, NVIDIA’s revenue totaled $215.94 billion, up 65% year-over-year; net profit was $120.08 billion, also up 65%.

The soaring revenue and high gross margins have made NVIDIA the king of AI monetization. At the same time, NVIDIA has become the foundation and ballast of AI infrastructure, driving the rise of the AI chip industry chain in recent years.

In the storage sector, Samsung and SK Hynix have shown significant growth. Since early 2025, Samsung’s stock price has nearly quadrupled, SK Hynix’s stock price has surged sixfold, and this has helped South Korea’s composite stock index hit a record high this week. Several analysts believe that South Korea’s chip leaders will continue to benefit from current memory supply tightness and strong AI demand, with Citigroup and Macquarie Capital both raising their target prices.

TSMC’s stock performance remains strong, far outperforming the broader market, with a 27.59% increase since early 2026, and a market value around $2 trillion. Additionally, leading companies in server manufacturing, packaging, and testing have also seen rising stock prices, creating a resonance effect centered on computing power demand.

Looking ahead to fiscal year 2027, NVIDIA’s guidance continues to grow, with an expected revenue of $78 billion for the first quarter, again surpassing analyst expectations. However, the Chinese market still holds uncertainties; the financial report notes that, so far, NVIDIA has not generated any revenue from the H200 licensing project.

NVIDIA’s unexpected growth has become the norm, and the capital markets are becoming more cautious. After such an impressive earnings report, NVIDIA’s stock price only rose by 1.41% on the same day.

NVIDIA founder and CEO Jensen Huang remains optimistic about AI’s prospects. He stated that computing demand is growing exponentially, the inflection point for Agentic AI has arrived, and enterprise adoption rates for agents are exploding. Customers are accelerating investments in AI computing infrastructure. Huang also revealed that NVIDIA is continuing efforts with OpenAI to reach cooperation agreements, “We believe we are close to an agreement.”

Data Center Surge

Focusing on core business lines, for the full fiscal year 2026, NVIDIA’s data center business contributed $193.48 billion in revenue, up 68% year-over-year, becoming the profit pillar.

In Q4, the data center segment performed best, with revenue reaching $62.3 billion, up 75% year-over-year and 22% quarter-over-quarter, accounting for over 91% of total company revenue. Breaking it down further, “compute business” (mainly GPU products) contributed $51.3 billion, up 58%; “network business” generated $11 billion, up 263%.

(NVIDIA Q4 data center revenue, Source: NVIDIA official website)

This indicates that both high-end GPUs used for large model training and high-speed network components supporting large clusters are in growth. In computing deployment, besides the well-known Blackwell series chips, NVIDIA’s interconnect products such as NVLink architecture, Ethernet, and InfiniBand platforms are also rapidly developing.

It’s clear that NVIDIA is accelerating the improvement of a complete AI infrastructure system. Currently, Blackwell chips remain NVIDIA’s main growth engine, including products like GB200 and GB300. The financial report shows that in the past four quarters, NVIDIA shipped 6 million Blackwell GPUs. Huang Huang previously stated in October last year that, over the next five quarters until 2026, the visible revenue from Blackwell and Rubin will exceed $500 billion, with orders around 20 million GPUs.

For the next-generation Rubin platform, NVIDIA says it will include six new chips, which can reduce inference token costs by up to 10 times compared to the Blackwell platform. AWS, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure will be among the first to deploy based on Vera Rubin.

NVIDIA CFO Colette Kress said earlier this week that NVIDIA has shipped the first samples of Vera Rubin to customers and plans to begin production and shipment in the second half of this year as scheduled.

Meanwhile, outside of data centers, other NVIDIA business units are also growing, with Blackwell being a key driver.

In the gaming segment, Q4 revenue was $3.7 billion, up 47%, though it decreased 13% quarter-over-quarter due to natural channel inventory adjustments. NVIDIA expects supply constraints to remain a challenge for the gaming business in Q1 2027 and beyond.

The professional visualization segment earned $1.3 billion in Q4, up 159% year-over-year and 74% quarter-over-quarter; automotive revenue was $604 million, up 6% year-over-year and 2% quarter-over-quarter.

Currently, NVIDIA’s market capitalization is about $4.7 trillion, roughly accounting for 3% of the global stock capital market. NVIDIA’s true strength lies in its ecosystem influence and its forward-looking leadership in supporting industries.

NVIDIA impacts the global tech industry through three levels: its chip product line (GPU platform), its technological platform (CUDA architecture), and the global AI ecosystem based on computing power.

What truly determines its market value is the outlook for the global AI ecosystem driven by computing power. With a market cap nearing $5 trillion, NVIDIA’s growth must match the expansion of global wealth.

Bubble or Future Outlook

Can the demand for computing power continue to grow? Is there a bubble ahead? These questions remain a focus of industry attention.

Kress said that since ChatGPT’s emergence, NVIDIA’s data center revenue has increased nearly 13-fold, with hyperscale cloud companies being the largest customers, accounting for over 50% of data center revenue.

Recent data shows that in 2026, the capital support from leading cloud providers continues to grow.

TrendForce’s data provided to 21st Century Business Herald shows that to accelerate AI application deployment and upgrades, global cloud service providers (CSPs) are increasing investments in AI servers and related infrastructure. It is estimated that in 2026, the combined capital expenditure of the top eight CSPs will exceed $710 billion, with an annual growth rate of about 61%. The eight top CSPs include Google, AWS, Meta, Microsoft, Oracle, Tencent, Alibaba, and Baidu.

TrendForce notes that, besides continuously purchasing NVIDIA and AMD GPUs, companies are also expanding ASIC infrastructure to ensure AI application service compatibility and cost efficiency in data center construction.

This means that on one hand, the demand and scale of the computing power market will continue to expand; on the other hand, NVIDIA faces increasing competition from other tech giants, with AMD and Meta signing large orders recently, and Google’s TPU also expanding its footprint.

Regarding growth concerns, Huang Huang said during the earnings call that he is confident in the cash flow growth of top clients. The significance of computing has changed. In the past, software ran on computers, with about $300-400 billion annually spent on purchasing medium-sized computers. Now, the demand has shifted to AI, where generating tokens requires massive computing power, directly translating into growth and revenue.

Huang Huang remains optimistic about the global data center capital expenditure reaching $3-4 trillion by 2030. “AI will not regress; AI will only get better.”

Meanwhile, as the era of AI inference accelerates, NVIDIA is also strengthening its architecture and inference capabilities. Huang Huang said NVIDIA will expand its architecture with Groq as an accelerator, similar to how Mellanox extended its own architecture, and more details will be shown at GTC.

Currently, NVIDIA remains the dominant force in AI, the absolute benchmark for global AI chip development, and a foundational enterprise for global technological progress. Recently, a “AI End-of-Day Report” triggered market panic, but NVIDIA’s performance and growth continue to reassure investors.

In fact, AI is not the first technology to cause job losses—historical stories like the textile machine era are well known. However, it is the first technology that has caused a widespread sense of unemployment risk, even within the tech industry itself. The controversy over textile machines impacting employment faded after years of development, leading to long-term growth in the textile industry.

Whether AI can create more productivity and truly achieve inclusive AI is a grander challenge that companies like NVIDIA need to consider.