NVIDIA's latest earnings report resembles a top student consistently excelling in class.

Revenue soared to $81.62 billion, marking an 85% year-over-year increase; data center revenue reached $75.2 billion, up 92% year-over-year; adjusted EPS stood at $1.87, a 140% year-over-year surge; gross margin hit 75%, exceeding market expectations; and next quarter's revenue guidance of $91 billion set another record.

For most companies, such results would be cause for celebration.

However, NVIDIA's market reaction was nuanced. Its after-hours stock price initially dropped more than 3%. The reason is straightforward: Wall Street no longer settles for NVIDIA merely "doing well"—it demands evidence that NVIDIA can continue to achieve extraordinary results.

This is what sets NVIDIA apart today: it's no longer just a typical tech company or a standard chip stock; it now functions as the general ledger for the entire AI capital expenditure cycle. Whenever NVIDIA releases earnings, the market scrutinizes not just its revenue but also the health of the global AI supply chain, cloud provider spending, and the growth prospects of sectors like HBM, CoWoS, optical modules, power, liquid cooling, and network chips.

My assessment is clear: NVIDIA's AI strategy has transitioned into the financial realization phase, moving beyond mere valuation narratives. Yet, the challenge lies in the fact that the more it delivers, the more discerning the market becomes.

I. AI: Beyond Hype, It's Now in the Income Statement

Many companies now enthusiastically discuss AI, suggesting it will revolutionize everything in the next decade. However, when you examine their earnings reports, revenues and profits remain largely unchanged—AI is merely a buzzword in their PowerPoint presentations.

NVIDIA is different.

The standout feature of this Q1 earnings report is that AI has directly translated into revenue and profit. Data center revenue of $75.2 billion accounts for over 90% of the company's total revenue. In essence, NVIDIA's growth is now primarily driven by AI computing demand.

More critically, there's differentiation within the data center segment.

Computing revenue reached $60.4 billion, up 77% year-over-year; networking revenue soared to $14.8 billion, a 199% year-over-year increase. These figures are significant. Previously, the market primarily focused on NVIDIA's GPUs. Now, while GPUs remain the core, networking, interconnects, switching, and system-level solutions are emerging as new profit drivers.

What does this signify?

NVIDIA is no longer just selling "a card" but an entire infrastructure for AI factories. Customers are not just purchasing GPU performance but also seeking training speed, inference efficiency, cluster stability, energy consumption, time-to-market, software ecosystem, and total cost of ownership.

Jensen Huang's key message during the earnings call was straightforward: customers are building AI factories, and the true metric isn't the price of a single GPU but how many tokens can be produced per watt and how much intelligence can be generated per dollar.

While this may sound like marketing, capital markets appreciate such rhetoric because it elevates NVIDIA beyond the chip cycle and positions it within an infrastructure return-on-investment framework.

Chip companies are subject to hardware cycles; AI factory platforms ride global capital expenditures. The valuation logic is fundamentally different.

II. NVIDIA's Real Strength: Expanding from GPUs to a Platform

I've always believed that many underestimate NVIDIA's ambition.

If NVIDIA were solely focused on selling GPUs, it would eventually face two pressures: cloud providers developing their own chips and competitors like AMD, Intel, Broadcom, and Marvell encroaching on market share.

However, NVIDIA is transforming itself into the "computing operating system" of the AI era.

The GPU is merely the entry point. Behind it lie CUDA, NVLink, InfiniBand, Spectrum-X, BlueField, Dynamo, Vera CPU, the Rubin platform, and an increasingly vital data center network.

During this earnings call, the Vera CPU emerged as a crucial new element. Management stated that Vera opens up a $200 billion market that NVIDIA hasn't fully tapped into before, with nearly $20 billion in CPU revenue visibility this year.

If this materializes, NVIDIA won't rely solely on GPUs for growth. It will add a CPU curve, on top of networking, systems, software, and edge AI.

I believe this is the most noteworthy AI signal for investors in this earnings report: NVIDIA is actively expanding its valuation anchor.

Previously, the market valued NVIDIA based on Blackwell shipments, HBM supply, CoWoS capacity, and cloud provider Capex. Now, it must also consider Vera Rubin, data center networking, AI Cloud, sovereign AI, enterprise AI, and physical AI.

NVIDIA has also revised its earnings disclosure framework, dividing its business into two major platforms: Data Center and Edge Computing, with Data Center further split into Hyperscale and ACIE.

This move isn't just about changing reporting metrics.

Hyperscale refers to mega-cloud customers like Microsoft, Amazon, Google, and Meta; ACIE includes AI Cloud, industrial AI, enterprise AI, and sovereign AI. NVIDIA is essentially signaling to the market: I'm not just reliant on a few large cloud providers for orders—my customer base is expanding.

This is crucial for valuation.

If NVIDIA were solely tied to a few large cloud customers, the market would eventually worry about Capex slowdowns. If Microsoft, Meta, or Google announces a reduction in capital spending, NVIDIA's valuation would suffer.

However, if ACIE continues to grow and AI cloud, enterprise, sovereign, and industrial customers start to dominate, NVIDIA's growth logic will shift from the "mega-cloud purchasing cycle" to the "global AI infrastructure expansion cycle."

These are two entirely different valuation models.

III. Strong Earnings, Yet Market Demands 'Greater Outperformance'

The most intriguing aspect now is that NVIDIA delivered a robust earnings report, yet its stock price didn't immediately soar.

This isn't because the market doesn't understand the report—it's because the market understands NVIDIA too well.

In the past, NVIDIA only needed to exceed expectations for funds to flood in. Not anymore. Because "NVIDIA exceeding expectations" has now become the consensus expectation.

Bank of America previously highlighted an interesting metric: over the past ten quarters, NVIDIA's actual revenue has averaged 7-8% higher than management guidance. Based on this historical pattern, Q1 revenue would need to reach $83-84 billion to truly surpass buyer expectations.

Instead, NVIDIA's actual revenue was $81.62 billion—higher than Wall Street's average expectation but below the most optimistic funds' threshold.

That's why the stock price fell after hours.

The market isn't criticizing the earnings report—it's saying: It's good, but not good enough to significantly revise my 2027-2028 profit models upward.

This is NVIDIA's current dilemma—and its strength. Ordinary companies rely on storytelling to repair valuations; NVIDIA has entered a phase where it must deliver exceptional performance every quarter.

The next quarter's revenue guidance of $91 billion carries the same implication. This figure is higher than the market average but below the most optimistic $96 billion expectation. The company also specifically noted that this guidance does not include revenue from Chinese data center computing.

This statement actually leaves room for adjustment.

If Chinese revenue recovers in the future, NVIDIA has room for additional upward revisions; if China continues to contribute nothing, it can still hit $91 billion with non-Chinese markets. For bulls, this signals insatiable demand; for cautious investors, it means Chinese uncertainty remains.

I believe that over the next two quarters, investors should focus not on whether NVIDIA will keep talking about AI but on several harder metrics.

First, whether Q2 actual revenue significantly exceeds the $91 billion guidance. NVIDIA doesn't just need to beat—it needs to beat by a substantial margin.

Second, whether gross margins can stay around 75%. With Blackwell ramp-up, Vera Rubin transition, and high costs for HBM and advanced packaging, maintaining margins would demonstrate NVIDIA's strong pricing power.

Third, whether data center networking can continue its high growth. Networking revenue up 199% year-over-year is the best evidence that AI factories are moving from "buying cards" to "buying systems."

Fourth, whether ACIE can continue to outperform. As long as AI Cloud, enterprise AI, and sovereign AI keep scaling, the market will believe NVIDIA isn't solely reliant on mega-cloud providers.

Fifth, whether Vera CPU can deliver as management expects. Nearly $20 billion in revenue visibility this year—if this materializes, it becomes NVIDIA's next growth curve.

Finally, keep an eye on Chinese revenue. If the H200 or other products regain sales access to China, it could serve as an additional catalyst.

Conclusion: NVIDIA: Not Just an AI Theme Stock, But the General Ledger for AI Capital Expenditures

This earnings report suggests that NVIDIA hasn't cooled down the AI rally—it's given the AI bull market another boost.

But this boost isn't for all AI stocks.

It's for the AI infrastructure chain: HBM, CoWoS, advanced packaging, optical modules, Ethernet switching, ASICs, liquid cooling, power, transformers, and data centers. Capital will likely shift from focusing solely on NVIDIA to spreading into these bottleneck assets.

This is my final assessment of NVIDIA's earnings report:

The greatest value of AI to NVIDIA isn't boosting its valuation but reshaping its revenue structure and growth model. Valuation is just the result—growth is the foundation.

NVIDIA has proven it isn't riding the AI hype wave alone but has truly transformed AI into revenue, profit, cash flow, and shareholder returns. The new $80 billion buyback and dividend increase from $0.01 to $0.25 aren't moves a company still burning money for stories would make.

But the market will now be even more realistic.

NVIDIA has shifted from "AI's biggest winner" to "AI cycle's final examiner." Every quarter, it must prove not just its own strength but that global AI capital expenditures are still rising, cloud providers are still spending, inference demand is still exploding, and networking and CPUs can open new markets.

So here's how I summarize NVIDIA's capital story now:

It is no longer just an AI theme stock but the most critical infrastructure tax collection point in the AI era. It's just that this entry point is too expensive, too crowded, and too highly anticipated—so every earnings report is no longer just about reporting results but about undergoing a market-wide audit.

Source: U.S. Stock Research Society