Quarterly earnings from Nvidia (NVDA.O) on Wednesday stand as a significant event for markets amid investor scrutiny regarding substantial spending in artificial intelligence (AI). Nvidia’s results could reassure investors questioning ongoing investments in AI technologies.
Wall Street on edge: Nvidia’s earnings could make or break the AI boom
Market estimates project Nvidia’s fourth-quarter sales at $38.5 billion, with first-quarter guidance around $42.5 billion. Options indicate a potential share price move of about 8% in either direction should the results deviate from expectations. “This earnings report isn’t just about Nvidia… It’s about whether the AI revolution can maintain its breakneck pace,” stated Jacob Falkencrone, global head of investment strategy at Saxo.
Q3 Fiscal 2025 Summary – GAAP (Millions of USD, except EPS):
| Metric | Q3 FY25 | Q2 FY25 | Q3 FY24 | Q/Q Change | Y/Y Change |
|---|---|---|---|---|---|
| Revenue | $35,082 | $30,040 | $18,120 | Up 17% | Up 94% |
| Gross Margin | 74.6% | 75.1% | 74.0% | Down 0.5 pts | Up 0.6 pts |
| Operating Expenses | $4,287 | $3,932 | $2,983 | Up 9% | Up 44% |
| Operating Income | $21,869 | $18,642 | $10,417 | Up 17% | Up 110% |
| Net Income | $19,309 | $16,599 | $9,243 | Up 16% | Up 109% |
| Diluted Earnings Per Share* | $0.78 | $0.67 | $0.37 | Up 16% | Up 111% |
| Metric | Q3 FY25 | Q2 FY25 | Q3 FY24 | Q/Q Change | Y/Y Change |
|---|---|---|---|---|---|
| Revenue | $35,082 | $30,040 | $18,120 | Up 17% | Up 94% |
| Gross Margin | 75.0% | 75.7% | 75.0% | Down 0.7 pts | — |
| Operating Expenses | $3,046 | $2,792 | $2,026 | Up 9% | Up 50% |
| Operating Income | $23,276 | $19,937 | $11,557 | Up 17% | Up 101% |
| Net Income | $20,010 | $16,952 | $10,020 | Up 18% | Up 100% |
| Diluted Earnings Per Share* | $0.81 | $0.68 | $0.40 | Up 19% | Up 103% |
Concerns regarding DeepSeek, a Chinese startup claiming to produce competitive AI technology at lower costs, have prompted increased scrutiny of major U.S. tech firms. However, Reuters sources report that Chinese companies have raised orders for Nvidia’s H20 chip, responding to growing demand for cost-effective AI models. DeepSeek is reportedly hastening the launch of its R2 model, which could heighten U.S. government apprehension over AI leadership, viewed as a national priority.
The technology tensions between the U.S. and China are expected to escalate, especially as the Trump administration prepares to implement stricter semiconductor export regulations and has already limited Chinese investments in strategic U.S. sectors. Amid these developments, Hong Kong-listed companies saw a 2.5% increase in shares, while technology stocks surged by 3.7% on Wednesday.
In the U.S., Treasury yields have rebounded following the advancement of Trump’s $4.5 trillion tax-cut plan, as investors remain cautious due to a deteriorating economic outlook and rising speculation about interest rate cuts from the Federal Reserve. Fed funds futures indicate more than 50 basis points of easing priced in by year-end, a substantial increase from the previous week’s estimates.
Nvidia is on track to complete a remarkable fiscal year, with analysts forecasting $38 billion in sales for the quarter ending in January, representing a 72% year-on-year growth. This January quarter marks the conclusion of a fiscal year in which Nvidia’s sales have more than doubled, primarily driven by the essential role of its data center GPUs in developing and deploying AI services, such as OpenAI’s ChatGPT. Over the past two years, Nvidia’s stock has appreciated by 478%, occasionally achieving a market capitalization exceeding $3 trillion.
However, Nvidia’s stock has decelerated recently as investors question the chipmaker’s trajectory amid signs that its key customers may be beginning to restrain spending following years of extensive capital expenditures. This wariness has been heightened by breakthroughs in AI technologies emerging from China.
A significant portion of Nvidia’s revenue is generated from major companies constructing extensive server farms, known as “hyperscalers,” which typically rent services to other firms. Reports from February indicated that a single customer accounted for 19% of Nvidia’s total revenue in fiscal 2024.
Recent estimates from Morgan Stanley suggest that Microsoft will constitute nearly 35% of spending in 2025 on Nvidia’s latest AI chip, Blackwell, with Google at 32.2%, Oracle at 7.4%, and Amazon at 6.2%. Consequently, any indication that Microsoft or its competitors might curb their spending could negatively impact Nvidia’s stock performance.
TD Cowen recently reported that Microsoft had canceled leases with private data center operators and revised its plans for international data centers in favor of domestic initiatives, sparking concerns about the sustainability of AI infrastructure growth and posing potential challenges for Nvidia’s chip demand. In response, Microsoft reaffirmed plans to invest $80 billion in infrastructure in 2025, asserting, “While we may strategically pace or adjust our infrastructure in some areas, we will continue to grow strongly in all regions.”
Over the past month, Nvidia’s primary customers have announced sizable investments, including Alphabet’s targeted $75 billion in capital expenditures for the year, Meta’s anticipated $65 billion, and Amazon’s goal of $100 billion. Analysts estimate that around half of AI infrastructure capital expenditures are captured by Nvidia.
Despite some hyperscalers exploring AMD’s GPUs and developing proprietary AI chips to reduce reliance on Nvidia, the corporation continues to dominate the cutting-edge AI chip market. Currently, Nvidia’s chips are primarily utilized for training advanced AI models, which entail significant costs. After the development phase by firms such as OpenAI, Google, and Anthropic, vast amounts of Nvidia GPUs are necessary to operationalize these models for consumer use, predicting continued revenue growth for Nvidia.
The emergence of DeepSeek last month introduced an efficient AI model that indicated top-tier Nvidia GPUs might not be required for training and utilizing cutting-edge AI, temporarily depressing Nvidia’s stock and resulting in a significant loss of nearly $600 billion in market capitalization.
Nvidia CEO Jensen Huang is set to address these concerns on Wednesday, focusing on the continued demand for GPU capacity in the AI sector. Huang has referred to a “scaling law” articulated by OpenAI in 2020, suggesting that AI models improve with increased data and compute usage. He has indicated that the DeepSeek model may represent a new aspect of this scaling law, termed “Test Time Scaling,” which posits that enhancing AI operational deployment will necessitate more GPU resources.
According to Bloomberg, Wall Street anticipates Nvidia will report earnings per share (EPS) of $0.84 on revenue of $38.2 billion for the quarter ending in January. While this reflects a 63% increase in EPS and a 73% rise in revenue compared to the previous year, investor expectations may render this growth underwhelming, especially considering last year’s figures of 486% EPS growth and a 265% revenue increase in Q4.
Nvidia’s data center segment is expected to yield the majority of its revenue, estimated at $34 billion, with the gaming sector contributing approximately $3 billion. The remaining revenue is projected to come from professional visualization, automotive, and OEM segments. Nvidia shares are down more than 5% year-to-date, reflective of challenges faced by major tech firms, including a nearly 7% decline for Alphabet, 3% for Amazon, over 5.5% for Microsoft, and over 1% for Apple. In contrast, Meta has experienced a better performance, with shares increasing by over 10%.
This earnings report is Nvidia’s first since the significant market impact caused by DeepSeek’s announcement and the subsequent $600 billion market cap loss in January. Concerns were raised when DeepSeek claimed functionality comparable to that of higher-end Nvidia chips, leading to fears regarding the viability of Nvidia’s more powerful processors.
In an interview with DDN CEO Alex Bouzari, Huang argued that utilizing high-performance chips for running models such as DeepSeek’s would actually drive demand for superior processors rather than diminish it. Nvidia also faces risks associated with potential tariffs on chips from Taiwan, as it collaborates with TSMC, which manufactures many of those chips. Trump has threatened potential tariff actions on Taiwanese imports and further restrictions on Nvidia chips supplied to China, jeopardizing revenue from significant markets.
While the majority of Nvidia’s revenue is generated from the U.S., where it earned $14.8 billion from $35 billion in sales during Q3, China remains an important market, contributing $5.4 billion and ranking third behind Singapore’s $7.6 billion intake.
Concerns have also been raised about the potential effects of Amazon, Google, Microsoft, and Meta developing custom AI chips that could rival Nvidia’s performance capabilities. If these custom solutions can achieve comparable results to Nvidia’s offerings, the demand for Nvidia chips may decrease significantly.
Cloud service providers, including Amazon, Google, and Microsoft, account for 50% of Nvidia’s data center revenue, meaning any loss in business from these firms would substantially impact Nvidia’s financial outcomes. However, Morgan Stanley Research analyst Joseph Moore advised caution against overreacting to such potentiality, drawing on past experiences with numerous Nvidia alternatives that initially garnered interest but ultimately fell short of establishing lasting market traction.
While Google and Amazon’s custom chips have shown potential, Moore asserts that Nvidia continues to expand its share within the AI domain.
Featured image credit: Nvidia
