NVIDIA: The undisputed AI infra King-will it surge more in 2025?

 

·         NVIDIA may report core EPS around $5.40 and $7.40 for Q3 & Q4FY26-TTM basis against market expectations (normal R/R) of $4.50 & $5.10

·         Overall, the AI infra market is expected to grow at a 40-50% CAGR to over $2T by 2030

·         NVIDIA CEO is ensuring a fine policy balance between Trump/US and China, ensuring no tech & trade war

Nvidia Corporation (NVIDIA), founded in 1993, is the undisputed global ‘King’ in the AI (Artificial Intelligence) ecosystem including GPUs (Graphics Processing Units), and SOCs (System-on-Chip Units). NVIDIA is engaged in designing and selling advanced computing solutions hardware and related software platforms, targeting several high-growth industries. NVIDIA is now virtually the global leader in design, development and marketing of programmable graphics processors, especially for the high-demand AI infrastructure ecosystem.

NVIDIA is now leading the global AI revolution. The market is optimistic about its current revenue growth above 100% and believes in future potential growth of around 50% for the next five years. Thus, NVIDIA's scrip is soaring amid the AI bandwagon, which the market still believes is not a bubble, like we have seen during the 2000’s dot-com frenzy. NVIDIA’s current market capitalization of ~4.3 trillion puts it equivalent to the world’s 4^th largest economy, Japan. NVIDIA’s current scrip price is around $177 and is expected to reach $205-225 mark by December’25, which would keep the nominal market capitalization in line with Germany (~$4.7T), at par with the 3^rd largest economy in the world!

NVIDIA’s diversified products & markets are positive for risk management:

Computing and networking solutions (78%)

·         Data center platforms and infrastructure, Ethernet interconnect solutions,

·         High-performance computing solutions, platforms, and solutions for enterprise AI infrastructure

·         Solutions for autonomous intelligent vehicles

·         Crypto-currency mining processors

·         Embedded computer boards for robotics, teaching, learning, and artificial intelligence development

GPU-Graphics processors Units and gaming PCs/accessories (22%)

·         Game consoles, video game streaming platforms, workstations (GeForce, NVIDIA RTX, Quadra)

·         Offers laptops, desktops, gaming computers, and computer peripherals (monitors, mice, joysticks, remote controls)

·         Software for visual and virtual computing, platforms for automotive infotainment systems, and cloud collaboration platforms

Revenues by major countries: US (44%); Taiwan (22%); China (17%); and other (17%)

Qualitative Analysis

Business Model and Strategic Positioning

Over the years, from a gaming GPU leader, NVIDIA has evolved into the backbone of the global AI ecosystem and has more than 90% of the AI ​​accelerator market. Its business model involves designing high-performance GPUs and AI platforms while outsourcing manufacturing to Taiwan’s TSMC. This helps lower CAPEX (capital expenditures) and higher Free Cash Flow (FCF) for NVIDIA. Also manufacturing outsourcing from Taiwan and selling into the US is resulting in a comparatively higher gross profit (GP) margin of around 76%, keeping competitors behind in the AI race in terms of quality & affordability.

The company’s strength lies in its integrated hardware-software stack: CUDA, Omniverse, and AI Enterprise create a sticky ecosystem, locking in developers and enterprises. Strategic moves in 2025—such as the $100 billion OpenAI investment, $5 billion Intel stake, and Alibaba robotics partnership, thanks to higher FCF—shift NVIDIA from a chip supplier to an AI “platform orchestrator,” securing long-term demand and revenue visibility and diversifying supply chains, derisking U.S.-China tech war tensions. Moreover, NVIDIA’s partnership with Hyperscalers like OpenAI, Microsoft, and automotive giants like Toyota ensures revenue visibility and a strong order book.

NVIDIA CEO Jensen Huang’s diplomatic balancing act is a unique asset

Huang’s ability to navigate U.S. and Chinese leadership—securing H20 chip sale approvals and fostering partnerships—has reopened a ~$20 billion annual revenue stream from China while aligning with Trump’s reshoring agenda. NVIDIA’s pivot to “physical AI” (robotics, autonomous vehicles) and sovereign AI (e.g., £2 billion UK investment) taps into trillion-dollar markets, positioning the company as a linchpin for national AI strategies and industrial automation. Gaming (~15% of revenue) remains a cash cow, with RTX and DLSS 4 innovations spilling over into AI simulation tools.

Jensen Huang’s AI rock-star status—evident in his Beijing fan base and White House access—drives NVIDIA’s strategic agility. His ability to secure policy wins (e.g., H20 sales to China) and forge equity-tied partnership/JV deals (OpenAI, Intel) showcases unmatched CEO leverage. However, his White House heavy lobbying and “bridge-building” between the U.S. and China invite scrutiny, with some critics labeling him a “neutral middleman” caught in superpower agendas.

In fact, Huang's good relations with Trump and China are hugely helping NVIDIA sustain its AI throne, ensuring favorable policy with market access. For investors, it's a bullish signal amid the trade war fog. Huang is unofficially acting as a tech czar, his diplomacy nimbly defusing U.S.-China flashpoints and sustaining AI's global flow. It's prevented all-out tech & trade war scenarios like total tech decoupling, benefiting everyone from developers to investors. But it's fragile; a Huawei breakthrough, another DeepSeek moment, or tariff escalation could expose the limits. Huang is very practical as he knows China controls the basic raw materials of any chip, while the US is a pioneer in tech. Under Trump 2.0, NVIDIA’s high-end AI chip may be used as an instrument of the political game of chicken by both the US and China, but the CEO Huang is smart enough to navigate this for the overall interest of the company.

NVIDIA’s Competitive Moat: Upcoming Rubin GPU may be a game changer in 2026

The AI dominance of NVIDIA comes from its technology innovation leadership. For example, Blackwell GPUs provide 40 times more performance than the older models, and also keep themselves far ahead of any competition from other giants like AMD and Intel. NVIDIA's CUDA (Compute Unified Device Architecture) and upcoming Rubin GPUs (2.5x to Blackwell) are vital to its AI and computing dominance. CUDA is the software foundation or OS that makes NVIDIA GPUs versatile powerhouses, while Rubin represents the next hardware leap in the company's GPU roadmap. This may help NVIDIA to consolidate its near monopoly in the AI computing space, with rivals like AMD's ROCM lagging in adoption. In brief, the upcoming Rubin GPU may be a game-changer for NVIDIA in 2026, fueling next-generation AI optimism.

AI ecosystem and overall market trends & potential risks

The AI infrastructure market is growing at around 40-50% CAGR and is projected to reach $2-3 trillion by 2030 globally, driven by hyperscaler buildouts and agentic AI (autonomous systems). NVIDIA is ideally positioned, but the shift from AI training to inference could compress margins, as inference requires less compute power. The overall semiconductor (chip) sector is expected to grow 20% in 2025, but NVIDIA is outpacing peers due to its AI dominance.

Tailwinds-Advantages       

NVIDIA’s GPU dominance gives it a significant edge—its chips/GPUs are 40x faster than predecessors, and competitors; AMD and Intel lag in performance and software. CUDA and NVIDIA’s software stack create a moat, making it hard for customers to switch, especially after the upcoming Rubin GPU launch in early 2026, which would be 2.5x faster than even the existing Blackwell GPU. Strategic investments in AI infra ecosystem, like OpenAI and Intel, lock in demand.

The pivot to “physical AI” (robotics, autonomous cars, systems) and next-generation sovereign AI opens new growth avenues, potentially almost monopolizing the multi-trillion-dollar market. High margins and stock buybacks keep investors happy, with stock up ~45% in the last year vs SPX-500 rally of 15%; i.e., NVIDIA outperformed the broader market, especially after Trump’s Liberation Day April 2 tariff announcements, which put an exception on Chip/related tech products.

Headwinds-Disadvantages

·         Geopolitical risks: U.S.-China trade/tech wars or perceived Taiwan disruptions could choke TSMC’s supply, impacting 100% of NVIDIA’s chips. Competition is heating up—AMD’s MI300 and Huawei’s Ascend chips target cost-conscious buyers, and falling GPU rental prices (~40% drops) signal potential price wars.

·         Macro risks like inflation/stagflation and higher data center energy costs or a capex/discretionary tech spending slowdown could dent growth. Overcapacity—if hyperscalers overbuild and inference demand lags—could crash GPU prices, hitting margins.

·         Trump policy uncertainty risks: TSMC's U.S. factories will likely raise NVIDIA chip costs by 5-10% for U.S.-produced units due to inherent fab expenses, with potential 15-20% spikes if tariffs bite Taiwan output.

·         However, subsidies, automation, and tariff exemptions mitigate this, making the net impact low single digits—far below early fears of 50%+ hikes. For NVIDIA, it's a net positive: Enhanced resilience supports AI demand (e.g., Blackwell ramps), and customers like hyperscalers pay premiums for "Made in USA" security.

·         Overall, this pressure short-term margins but bolsters long-term dominance in a $2T AI market. NVIDIA’s aggressive investments, while strategic, strain cash flow and carry risk if partners underperform.

·         In addition, the potential changes for an estimation-centric/inference-focused AI vs training can compress revenue & margin if the demand for high-end GPUs/Chips softens.

In brief, NVIDIA’s operating model is a powerhouse for the AI ecosystem, which combines sophisticated hardware, state-of-the-art software, and solid investment to stay ahead in the global AI ​​race. It is well deployed for growth, but also faces risks from Trump’s uncertain policies (whims & fancies, morning mood & Truths), geopolitical fragmentations, competitions, and regulatory headwinds. Although AI is a great tool to improve productivity, it’s also a real threat to job losses, layoffs, higher unemployment/underemployment and social unrest.

The global AI market's potential 50% CAGR perception may be echoing a pragmatic shift to agentic and physical AI, with total economic impact potentially exceeding $15 trillion by 2030 globally. The US leads in innovation, the EU in regulation, and China in scale, can produce the same standard of US generative AI with only 25% of US costs as we have seen in recent DeepSeek disruption. Trump now wants to be far ahead of China in terms of the AI race without seeking decoupling, but ensuring strategic derisking. Thus, NVIDIA has a good opportunity to cash in on the AI revolution both in the US and China, along with the EU, controlling almost 75% of global economic output.

Investment Thesis: NVIDIA is a must for any Wall Street portfolio

One can argue about the time & price levels (technicals) for buying NVIDIA scrips, but can’t ignore the investment compulsion; it’s a must buy for any Wall Street portfolio. NVIDIA is now a fundamental pick for any AI-themed portfolio. The thesis boils down to NVIDIA's unmatched moat in AI hardware and software, coupled with explosive secular tailwinds in generative AI, agentic systems, and physical AI applications. With over 90% market share in AI accelerators, NVIDIA's CUDA ecosystem creates a lock-in effect that's hard for rivals to crack—developers build on it, enterprises standardize around it, and hyperscalers like OpenAI can't scale without it.

Quantitative Analysis-NVIDIA

Financial Performance FY25 Summary-NVIDIA’s financials reflect its AI-driven dominance:

·         Revenue: $130.5B vs 60.9 in FY24 (+114.2%); 5Y R/R: 117.4%

·         Gross Profit Margin: 75.0% vs 72.7% (Y/Y) and 63.3% in FY21; among the highest in tech peers

·         Core/EBTDA margin 63.7% vs 56.2% (y/y) and 33.8% in FY21

·         Core EPS $3.3 vs 1.4 (Y/Y) and 0.7 in FY21; 5Y R/R: 196.2%

·         Data Center contributes ~$112B; i.e., almost 85% of revenue, followed by Gaming (~$18B), and autonomous Automotive (~$2B).

Financial statement analysis & trend: FY25

Financial statement analysis & trend: Q2FY26-TTM

Financial Performance Q2FY26: TTM summary

·         Revenue: $165.2 billion vs 148.5 sequentially (+11.2%) and 96.3 yearly (+71.6%), driven by Blackwell GPU

·         Gross Margin: 69.8% vs 70.1% sequentially and 76.0% yearly

·         Core/EBTDA margin 59.3% vs 59.3% sequentially and 63.3% yearly

·         Core EPS $4.0 vs $3.6 sequentially (+11.8%) and 2.4 yearly (+62.5%); effective average R/R 13.4% sequentially

Management Guidance:

Guidance for FY26: Issued after FY25 earnings release

·         Revenue: $190-200B, to be led by data center and Blackwell Ultra

·         EBITDA: $120-130B at around 62–65% margins.

·         EPS: around $18.10 (non-GAAP), up400% from FY2023, supported by $60B share buy back

·         H20 Impact: $650M in Q2 sales (non-China) boosted margins and EPS, but China restrictions cap growth, which may ease in Q3

NVIDIA issued its latest official guidance on August 27, 2025, during Q2FY26 earnings release and concall. The management was quite confident about solid demand for Blackwell GPUs amid resilient demand for AI infra and a potential upside from China export approval, although they excluded Chinese revenue from its core guidance.

 Q3 FY2026 Guidance (August 2025 - October 2025)

·         Expected Revenue: $54B (±2%).

·         This translates to ~$223B TTM revenue for Q3FY26.

·         Expected Gross Margin: 72.5% (±0.5%), slightly lower due to Blackwell ramp-up costs

·         Expected Operating Expenses: $5.3B (±2%).

·         Revenue projection excludes $2-5B potential China H20 sales, which could boost revenue if licenses GET clear.

NVIDIA Q2 Guidance excludes any contributions from H20 AI chip shipments to China due to ongoing U.S. export uncertainties. Management highlighted a potential $2-5B upside if licenses are granted, but stressed this is "optional" and not factored in. CEO Jensen Huang reiterated "incredible demand" for Blackwell, with full production ramps underway and the Rubin platform teased for 2026. Data center remains the primary driver at around 80% of revenue on average.

FY27-28 (Assumptions)

·         Revenue: 30–40% CAGR.

·         EBITDA margins 60–65%

·         EPS: 40–50% CAGR

Usually, like most of the other companies, NVIDIA also guides by the next quarter during the quarterly earnings concall. However, management offered directional commentary and segment-specific insights, signaling sustained high growth for 2026-28:

·         Revenue growth 30-40% Y/Y in the longer run based on overall trends.

·         Management noted revenue growth in H2FY26 will remain above 50%

·         Gross Margin: ~75% for FY26 led by premium products & service like Blackwell and CUDA/Omniserve

·         Operating Expenses are expected to grow by ~30% (Y/Y), implying $18-19B; focused on AI talent, data center expansions, and sovereign AI initiatives

In brief, NVIDIA management sees strong demand for compute and networking, with Blackwell Data Center revenue already up 17% sequentially in Q2. Potential China H20 recovery ($4-5B quarterly if approved) could add 5-10% to full-year totals. Management highlighted "record" networking revenue in Q2 and teased Rubin GPUs for late FY2026/early FY2027. NVIDIA's guidance reflects confidence in the potential $2T AI infrastructure market by 2030, with the company capturing ~90% of AI accelerator share.

NVIDIA H20 Chip Sales to China: Approval Status from the Trump Admin

NVIDIA already obtained approval from the Trump administration to sell H20 AI chips to China, effective from July 2025. This is a major policy win for NVIDIA, unlocking the potential $20B Chinese market, which may boost H2FY26 topline and bottom line. CEO Jensen Huang’s diplomatic maneuvering secured this reversal, aligning with U.S. economic interests while navigating trade tensions. Huang convinced the Trump admin that H20 is no longer an ultra-advanced AI Chip and China may already be on the way to make the equivalent as per their overall policy not to depend on the US/any other country for critical techs (strategic derisking). However, China’s caution on H20 risks and potential new restrictions pose challenges. This approval strengthens NVIDIA’s AI dominance but requires vigilance for geopolitical shifts, with Q3 earnings & commentaries (November 2025) remaining key.

Potential Growth Drivers:

·         AI infrastructure spending ($3–4T by 2030) @40-50% CAGR

·         Blackwell Ultra and Rubin architectures will drive data center growth (50% of Q2 data center revenue from cloud providers).

·         Automotive and professional visualization growth (69% and 32% year-over-year in Q2).

Risk for outlook:

·         Geopolitics: U.S.-China export restrictions could reduce revenue by $5-10B annually, particularly for H20 and B30A chips.

·         Competition: AMD, Intel, and especially Chinese firms (Huawei, Cambricon) may erode NVIDIA’s near-global monopoly of 80% market share by offering a lower price and almost the same GPU speed

·         Supply Chain issues: Production constraints at TSMC/Samsung and H20 supply halts (August 2025) could limit output and revenue

·         Economic: Macroeconomic downturns as a result of Trump’s bellicose policies may reduce gaming and enterprise discretionary tech/AI spending.

·         H20 Context: The $180 million H20 inventory release in Q2FY26 mitigated prior $4.5B write-down charges, but ongoing China restrictions/curbs limit upside; while potential B30A/RTX6000D chips could offset losses.

Peer Comparison: Lowe PEG ratio indicates NVIDIA is still undervalued compared to peers and the overall market

·         NVIDIA’s premium valuation (45 PE) reflects AI innovation, optimism, scarcity, and huge EPS growth potential premium

·         Moderate PEG Ratio: Average projected PE 45 over 2026-28 against  estimated 50% EPS growth, suggesting 0.9 PEG and undervaluation for growth over peers and overall market trend; SPX-500 PEG is almost 2.5 on average

·         Market-Analyst Price Targets: $205-$235 for 2025

Fundamental Valuation & Outlook: NVIDIA

Considering the overall trend, run rate, pros & cons as discussed above, and NVIDIA’s latest guidance:

·         Core Revenue grew 114%; average R/R 117%; projected to grow ~131% to $301B in FY26

·         Core operating margin (EBTA) projected for FY26: 60.5% vs 59.9% actual Q2CY26 TTM and 63.8% in FY25

·         Core EPS projected for FY26: $7.40 vs 3.30 for FY25 (+120%) and 4.00 for Q2FY26-TTM

·         Core EPS grew +144% in FY25, and average R/R for the last 5 years was 196%; abnormally high growth due to a small base and growing business/revenue.

·         Core EPS for FY27 is projected to grow around 40% to $10.30 as per NVIDIA management guidance, and grow a stable base.

·         Estimated P/E: Worst-Base-Best-Bubble case: 20-35-45-55 for FY26-27 against average EPS growth 80% projected in this period and future potential growth 40-50% for the next few years (in line with overall AI market growth projections of 40-50%)

·         The average targets for FY26-27: $286-400

·         As the stock market usually discounts 1Y earnings in advance, the potential targets for June’26 and June’27 should be around $286-400

·         Short-term target $225-245 by December’25

·         As per the QTR TTM trend, the market is expecting core EPS $4.50 and $5.10 in Q3FY26 and Q4FY26 TTM.

·         But as per management guidance, core EPS may come around $5.40 and $7.40 for Q3 & Q4FY26.

Technical Outlook (LTP: 178 on 24/09/25): NVIDIA

·         Short & Mid-term support: 170/163-158/150 & 130/119

·         Long-term support: 110/100-90/85

·         Short & Mid Term resistance: 185/195-201/205 & 215/232

·         Long-term resistance: 255/279-319/355-375/425

Short-term swing trading idea:

·         Buy on dips around: 168-158; TGT: 185/195-200/205; TSL: 155

·         Sell on rise around: 185/195-200/205; TGT: 168-158; TSL: 208

The inverse H&S pattern formed in the weekly chart indicates a potential target of 225 by December’25. Also, a golden cross (short-term moving averages crossed above long-term) in the daily chart indicates strong bullish momentum, and buyers are in control.

Short/Midterm Investment idea: Buy on dips around 168-158; TGT: 225-245

Conclusion:

NVIDIA stands at an inflection point in global AI infra space, with a robust business model driving exceptional growth across all key markets. NVIDIA’s ability to sustain 40-50% CAGR in core EPS over the next 2-5 years (2026-2028/30) depends on its capacity to maintain innovation & dominance in AI, expand into new markets, and navigate risks. NVIDIA’s EPS growth has been stellar, fueled by explosive AI demand. Several tailwinds could support sustained EPS growth as the world is shifting now from the Google search age to generative & physical AI, which is also improving overall productivity and economic growth without fueling inflation. But at the same time, the shift to inference-focused AI vs training) could compress revenue & margins if demand for high-end AI GPUs/chips softens.