---
title: "NVIDIA Q1 Earnings: Behind the Fourth Post-Market Decline, Three Growth Curves Are Converging"
type: "News"
locale: "en"
url: "https://longbridge.com/en/news/287178662.md"
description: "Revenue of $81.6 billion, up 85% year-over-year, with data center revenue nearly doubling; Vera CPU opens a new $200 billion track, where the true expectation gap lies"
datetime: "2026-05-21T07:56:41.000Z"
locales:
  - [zh-CN](https://longbridge.com/zh-CN/news/287178662.md)
  - [en](https://longbridge.com/en/news/287178662.md)
  - [zh-HK](https://longbridge.com/zh-HK/news/287178662.md)
---

# NVIDIA Q1 Earnings: Behind the Fourth Post-Market Decline, Three Growth Curves Are Converging

## Key Takeaways

① Earnings beat expectations, yet the stock fell for the fourth consecutive time post-market: Non-GAAP EPS was $1.87, beating the consensus of $1.76 by 6%; Q2 revenue guidance of $91 billion significantly exceeded the expected $86.8 billion. However, NVDA dropped to $220.40 (-1.31%) in after-hours trading, with an average same-day return of -1.54% over the past four fiscal quarters—this is normal digestion under "perfect pricing," not a signal of weakening fundamentals.

② Data center revenue nearly doubled, with a new framework revealing more important structural differentiation: Data center revenue reached $75.2 billion, up 92.3% year-over-year. Starting this quarter, it is split into two sub-segments: Hyperscale ($38 billion, QoQ +12%) and ACIE ($37 billion, QoQ +31%). The growth rate of ACIE, driven by AI Cloud and Sovereign AI, is already 2.6 times that of Hyperscale, indicating rapid diversification of NVIDIA's customer base.

③ Revenue increased from $44.1 billion to $81.6 billion over five quarters, entering a stable trajectory of ~20% quarter-over-quarter growth: Since the mass production of Blackwell began in Q3 FY2026, the quarter-over-quarter growth rates for Q3/Q4 FY2026 and Q1 FY2027 were +22%/+20%/+20% respectively, demonstrating a flywheel effect on the demand side rather than a one-time surge.

④ **Vera CPU opens a brand new $200 billion TAM, with nearly $20 billion in revenue visibility this year:** NVIDIA previously held almost zero market share in the CPU sector. Vera CPU represents a genuine "incremental new market," forming the next-generation platform together with the Vera Rubin GPU, with inference throughput improved by 35 times compared to Blackwell. This signal has been almost entirely excluded from analyst models.

⑤ Non-GAAP figures are the correct focal point; GAAP net income of $58.3 billion includes a one-time investment gain of $15.9 billion: Non-GAAP net profit was $45.5 billion (+138.5% YoY), free cash flow was $48.6 billion, and FCF margin was 59.5%. The quarterly dividend was significantly raised from $0.01 to $0.25, and a new $80 billion share repurchase authorization was approved, marking unprecedented capital return intensity.

**⑥ Chinese revenue excluded from guidance due to policy impacts is an upside option rather than a base forecast:** Management explicitly stated that the current quarter's guidance does not include any data center computing revenue from China, with uncertainty remaining. Before policies become clear, potential revenue from the Chinese market is more of an upside option than a base forecast.

NVIDIA delivered a triple beat in its Q1 report: current financial figures exceeded expectations, Q2 guidance surpassed expectations, and strategic increments (Vera CPU) completely caught the market off guard. Historically, the company has entered a stable trajectory of ~20% quarter-over-quarter growth since Q3 FY2026, which is the financial imprint of the Blackwell super cycle materializing. Against the backdrop of accelerating global AI infrastructure spending, there has been no slowdown on the demand side—the capital expenditure flywheel of hyperscale customers continues to spin, while the growth rate of emerging customers in Sovereign AI and AI Cloud has surpassed traditional giants.

The post-market decline in stock price reflects valuation gaming rather than fundamental logic. While analysts generally raised their target prices to the $285-$325 range, **"the beat wasn't large enough" became the main cause of short-term sentiment suppression—this is essentially a "perfectness tax," an inherent attribute of highly priced assets. The consensus narrative shifted somewhat after the earnings report: from "whether Blackwell can be mass-produced on schedule" to "the ramp-up slope of Vera Rubin" and "the realization of Vera CPU revenue," with forward-looking speculation points switching to the next-generation platform.**

**The two most critical windows for observing NVIDIA are: first, whether Vera Rubin mass production proceeds as scheduled in Q3 this year and whether its ramp-up slope surpasses Blackwell; second, whether the $20 billion revenue visibility for Vera CPU within the year can turn into actual booked revenue.** The realization of these two variables will determine whether the current pricing of 50-55x NTM P/E is sustainable.

## I. Why the Post-Market Decline: The "Perfectness Tax" Replayed for the Fourth Time

After the close on May 20, NVIDIA announced its Q1 FY2027 earnings for the period ended April 26: revenue of $81.6 billion, up 85% year-over-year; Non-GAAP EPS of $1.87, beating the Wall Street consensus of $1.76 by about 6.25%; and Q2 revenue guidance of $91 billion, significantly exceeding the market expectation of $86.8 billion.

However, the stock fell in after-hours trading. NVDA closed at $223.33 (+1.23%) during regular hours, and was quoted at $220.40 at 5:14 PM EDT in after-hours trading, a drop of -1.31%. This marks the fourth consecutive time in the past four fiscal quarters that NVIDIA has reported "earnings beats but post-market declines," with an average same-day return of -1.54% across these four instances.

Key judgment: This is not a fundamental issue—rather, NVIDIA has entered a rare situation where expectations are so high they are nearly perfect, and any result that is merely "excellent" triggers profit-taking. The post-market decline is the "perfectness tax" levied by the market for a perfect delivery.

## II. From $44.1 Billion to $81.6 Billion: The Compounding Effect of Blackwell Mass Production

From $44.1 billion in Q1 FY2026 to $81.6 billion this quarter, NVIDIA nearly doubled its revenue over five fiscal quarters. More importantly, consider the growth slope: after Q3 FY26, NVIDIA entered a stable trajectory of "approximately 20% quarter-over-quarter growth each quarter"—the compounding effect brought by Blackwell mass production.

Earnings quality is equally outstanding: Non-GAAP gross margin was 75.0%, flat with the previous quarter and far higher than the 60.8% from a year ago (an increase of 14.2 percentage points); free cash flow was $48.6 billion, with an FCF margin as high as 59.5%—among the world's largest companies by market cap, this margin rivals the best software companies.

## III. Data Center Breakdown: Divergence and Convergence of Two Sub-Curves

Data center revenue was $75.2 billion, accounting for 92% of total revenue, up 92.3% year-over-year. This quarter, NVIDIA adopted a new framework for the first time, splitting the data center segment into two sub-segments:

**Hyperscale: Mature Flywheel Continues to Spin**

Public cloud and the world's largest consumer internet companies generated revenue of $38 billion, up 12% quarter-over-quarter. The capital expenditure flywheel of Google, Microsoft, Amazon, and Meta continues, with management expecting hyperscale capital expenditure to reach $1 trillion in 2027. Demand in this segment is fully priced in.

**ACIE (AI Cloud + Industry + Enterprise): Emerging Forces Surpass Traditional Giants**

AI Cloud + Industry + Enterprise + Sovereign AI generated revenue of $37 billion, up 31% quarter-over-quarter—nearly three times that of Hyperscale. Among this, AI Cloud revenue more than doubled year-over-year, and Sovereign AI (national-level AI infrastructure led by various governments) grew by over 80% year-over-year.

The absolute volumes of the two sub-segments are nearly equal ($38 billion vs. $37 billion), but the growth divergence is significant. NVIDIA's customer base is rapidly diversifying—this is the most powerful hedge against the structural risk of "computing power demand being highly concentrated among a few hyperscale customers."

On the product level, Networking revenue was $14.8 billion, up 199% year-over-year and 35% quarter-over-quarter, with growth far exceeding the Compute segment, reflecting explosive demand for interconnect infrastructure during the upgrade of AI training and inference architectures.

## IV. GAAP vs. Non-GAAP: A Number That Must Be Clarified

GAAP net profit was $58.3 billion, up 210.6% year-over-year—this figure was significantly inflated by a one-time equity securities investment gain of $15.9 billion, which is non-operating income. Excluding this item, the true operating net profit corresponds to the Non-GAAP figure of $45.5 billion, up 138.5% year-over-year. Using GAAP figures to tell NVIDIA's profitability story is a common trap.

In terms of capital returns, management conveyed confidence through action: Q1 saw a total of $19.3 billion in share buybacks plus $240 million in dividends, totaling $20 billion; the board approved a new $80 billion share repurchase authorization; the quarterly dividend was significantly raised from $0.01 per share to $0.25 (a 2400% increase); and 50% of FY2027's full-year FCF will be used for shareholder returns.

## V. Vera CPU: The Third Curve Outside Analyst Models

If the data center beat is "Expected Upside," then the emergence of Vera CPU is "Unexpected New TAM." NVIDIA's revenue visibility in the CPU sector this year is close to $20 billion (on a standalone CPU basis), whereas previously NVIDIA held almost zero market share in the CPU market.

CEO Jensen Huang announced on the conference call that Vera CPU opens a brand new $200 billion TAM. Vera is NVIDIA's self-developed 88-core CPU, which together with the Rubin R100 GPU forms the next-generation AI computing platform, Vera Rubin. Compared to the current Blackwell platform, Vera Rubin improves inference throughput by 35 times, reduces cost per token by 60%, and achieves a comprehensive revenue benefit for AI factories that is 10 times that of Blackwell.

The cumulative revenue expectation for the Blackwell+Rubin platform from 2025-2027 has reached $1 trillion (management guidance, including GPU racks). Adding the $20 billion from Vera CPU, NVIDIA is expanding its TAM definition from "AI accelerated computing" to "full-stack AI infrastructure."

## VI. China and Custom Chips: Two Clouds Hanging Overhead

Affected by relevant policies, management explicitly stated that the current quarter's guidance does not include any data center computing revenue from China, with related uncertainties still existing.

Before policy certainty becomes clear, potential revenue from the Chinese market is more of an "upside option" for NVIDIA rather than a base forecast. The true catalyst window lies in the first confirmation of Vera Rubin mass production this autumn.

**Custom Chip (ASIC) Threat Assessment:** Google TPU, Amazon Trainium, and Microsoft Maia series are all accelerating. Management directly addressed this concern on the conference call, stating that custom chips are "niche products" and that NVIDIA's inference market share is growing rapidly. Independent judgment: Custom chips have cost advantages for specific workloads of specific large customers (especially recommendation systems and large-scale inference), but the moat of generality and software ecosystem (CUDA) remains difficult to replicate within 3-5 years. The real threat window will only become clear with the next generation after Vera Rubin (around 2028).

## VII. Q2 Guidance and Analyst Target Prices

**Q2 FY2027 Guidance:** Revenue of $91 billion (±2%), exceeding the market expectation of $86.8 billion by about +4.8%; Non-GAAP gross margin of 75.0% (±50bps); GAAP operating expenses of approximately $8.5 billion, Non-GAAP approximately $8.3 billion; full-year operating expense growth maintains the "high 40s%" range, mainly driven by R&D and AI tool usage.

On the supply side, total supply (inventory + purchase commitments + prepayments) has reached $145 billion—management voluntarily disclosed this figure, sending the signal that "the constraint is the time window for meeting demand, not demand itself."

**Mainstream Institution Target Price Range:** HSBC $325, DA Davidson $300, Morgan Stanley $285, Wedbush $300. Corresponding to the closing price of $223.33 on the day, the average implied upside space is about 30%-46%. Analysts generally list the confirmation of Vera Rubin mass production and the realization of Vera CPU revenue as the catalysts for the next round of revaluation.

## VIII. Conclusion: The Dialectic of Pricing and Value

NVIDIA's four consecutive post-market declines prove one thing with data: when a company is "perfectly priced," the difficulty of beating expectations rises exponentially. But this is different from a "valuation bubble."

Based on Non-GAAP net profit of $45.5 billion and FCF of $48.6 billion, the current market cap corresponds to approximately 50-55x NTM P/E—high for tech growth stocks, but not outrageous. More critically, the third of the three growth curves (Vera CPU) has been almost entirely excluded from market modeling, while the acceleration of the second curve (ACIE/Sovereign AI) has just come into institutional view.

NVIDIA lacks neither demand, nor cash, nor pricing power. The only thing it needs to prove is execution—whether Vera Rubin can be mass-produced on schedule in Q3, whether its ramp-up slope surpasses Blackwell, and whether CPU revenue can realize the nearly $20 billion visibility this year.

The post-market decline is the "perfectness tax" levied by the market for a perfect delivery. And the next true catalyst is likely not in the next quarter's earnings report, but in the first confirmation of Vera Rubin mass production this autumn.

_Data sources: NVIDIA Q1 FY2027 Earnings Report (May 20, 2026), NVIDIA Earnings Conference Call Transcript, WebSearch Analyst Data Summary. GAAP net profit of $58.3 billion includes non-operating equity securities investment gains of $15.9 billion; the Non-GAAP figure of $45.5 billion is the correct reference benchmark for operating profit. This article is for reference only and does not constitute any investment advice._

### Related Stocks

- [NVDA.US](https://longbridge.com/en/quote/NVDA.US.md)
- [NVDU.US](https://longbridge.com/en/quote/NVDU.US.md)
- [NVDY.US](https://longbridge.com/en/quote/NVDY.US.md)
- [NVDL.US](https://longbridge.com/en/quote/NVDL.US.md)
- [NVDX.US](https://longbridge.com/en/quote/NVDX.US.md)
- [07788.HK](https://longbridge.com/en/quote/07788.HK.md)
- [07388.HK](https://longbridge.com/en/quote/07388.HK.md)
- [NVDD.US](https://longbridge.com/en/quote/NVDD.US.md)
- [NVDQ.US](https://longbridge.com/en/quote/NVDQ.US.md)
- [SOXL.US](https://longbridge.com/en/quote/SOXL.US.md)

## Related News & Research

- [Research Alert: CFRA Maintains Strong Buy Opinion On Shares Of Nvidia Corporation](https://longbridge.com/en/news/287202154.md)
- [Nvidia posts record $81.6B quarter, unveils Vera CPU](https://longbridge.com/en/news/287165782.md)
- [Nvidia Will Report Q1 Earnings on May 20 — Options Traders Expect an 8.65% Move in NVDA Stock](https://longbridge.com/en/news/286763191.md)
- [Nvidia earnings on May 20 could sway AI market sentiment](https://longbridge.com/en/news/286642553.md)
- [Up 110% YTD, Why Nvidia’s Backing Could Keep Nokia Stock Rallying](https://longbridge.com/en/news/287094557.md)