NVDA

RED FLAG: ABSENT

The balance sheet does not present fatal risks. Cash and short-term investments of $62.56 billion amply exceed gross debt ($8.47 billion), FY2026 cash from operations was $102.72 billion and cash generation makes the company structurally solvent under any realistic stress scenario. Relevant industrial commitments (manufacturing commitments $95.20 billion, cloud $27.00 billion) are monitored but do not constitute a structural balance sheet risk.

AI DISRUPTION RISK: LOW

NVIDIA is the primary infrastructure provider of the AI revolution, not one of its targets. The company asymmetrically gains value as artificial intelligence adoption grows. The risk of substitution from its own core business is structurally low in the short to medium term, given the depth of the CUDA ecosystem and the industry's dependence on the entire proprietary hardware-software stack.

B1.1 — Leadership and systemic role: 9.50

NVIDIA holds more than 85% market share in GPU accelerators for artificial intelligence training in global data centers. Compute & Networking revenue in FY2026 reached $193.48 billion out of a total of $215.94 billion (+65% YoY), consolidating a position of systemic centrality that no competitor is able to challenge in the short term. Q4 FY2026 recorded record revenue of $68.13 billion (+73% YoY), with Data Center alone at $62.30 billion (+75% YoY). Hyperscalers such as Microsoft, Google, Amazon and Meta depend on NVIDIA GPUs to develop their own foundational models, making the company a quasi-critical infrastructure of the global digital transition.

B1.2 — Customers and barriers to entry: 9.25

The competitive moat does not lie only in hardware: the CUDA ecosystem — the parallel programming framework on which millions of developers have built their code over the past fifteen years — creates a prohibitive switching cost toward alternative architectures. Added to this is control over the full stack: chips, networking (Mellanox), systems, software libraries and SDKs. Customer concentration is significant (the main hyperscalers account for a relevant portion of revenue), but this concentration reflects the fact that the players with the greatest computing resources in the world have chosen NVIDIA as the standard. The development of custom silicon by customers (Google TPU, Amazon Trainium, Meta MTIA) is already incorporated into the score: the moat is exceptional but not absolute over a multi-year horizon.

B1.3 — Business economics: 9.75

The economics of the business are exceptional for a company classified as a hardware manufacturer. FY2026 GAAP gross margin stood at 71.1%, with operating income of $130.39 billion and net income of $120.07 billion (net margin 55.6%). ROIC exceeds 126%, a level normally associated with software businesses with patents, not fabless semiconductor producers. This structure reflects the pricing power derived from the absence of comparable alternatives in the market.

B1.4 — Balance sheet and resilience: 9.00

The financial position is very solid: cash and short-term investments of $62.56 billion against gross debt of $8.47 billion (Debt/Equity 7.25%). FY2026 operating cash generation of $102.72 billion offers enormous room for maneuver; TTM FCF is $96.68 billion. Shareholder returns in FY2026 were $41.10 billion between buybacks and dividends, with residual authorization of $58.50 billion. The score does not reach the maximum due to dependence on TSMC for physical production and exceptional-scale industrial commitments, which require continuity of demand.

B2.1 — Sector cycle: 8.90

The advanced semiconductor cycle is clearly expansionary and accelerating compared with the previous analysis. Gartner forecasts global semiconductor revenue above $1.30 trillion in 2026 (+64%), with worldwide IT spending at $6.31 trillion (+13.5%), driven by AI infrastructure and software. Sector earnings revisions continue to be positive, B2B demand structurally exceeds supply, and credit stress is absent due to the massive flow of capital into AI infrastructure. The cycle does not reach the maximum because elements of latent volatility remain linked to potential capex rationalization by hyperscalers.

B2.2 — Structural trends: 9.50

The macro trend is among the most powerful identifiable in the current decade. Demand for AI computing for training, inference, agentic AI and rack-scale systems is structural and multi-decade. Computational sovereignty — nation states building independent AI clusters — creates a new and stable class of buyers. Morgan Stanley identifies in agentic AI an expansion of chip demand well beyond GPUs, toward CPUs and other accelerators, further expanding the TAM of the NVIDIA ecosystem. The tailwind is clear and durable.

B2.3 — Competitive positioning in the cycle: 9.50

In the current cycle NVIDIA captures asymmetrically most of the economic value created by the AI infrastructure boom. FY2026 Data Center reached $193.70 billion (+68% YoY), and the architectural roadmap — from Hopper to Blackwell to Vera Rubin — maintains a structural technological lead over competitors. AMD is advancing in inference, but it does not replicate the software ecosystem. Hyperscaler custom silicon is real but concerns specific workloads, not general-purpose training where NVIDIA is irreplaceable.

B2.4 — Specific exogenous risks: 5.50

Exogenous risks remain real and have not improved compared with March. The tightening of export restrictions toward China (H20 ban) has further limited a historically relevant market, with estimated impacts above $15 billion in potential revenue. Dependence on TSMC for physical production represents a structural supply chain risk: an interruption of Taiwanese foundries would have a devastating impact on the entire Western supply chain. The score reflects real, documented and partly already materialized risks.

B3.1 — Intrinsic Fair Value: 6.38

The four DCF sources present very divergent estimates, reflecting the extreme sensitivity of models to terminal growth assumptions for a company with this profile.

The weighted FV of $209.14 (equal weights of 25% for GROWTH type) indicates that at $199.88 the stock trades at a 4.4% discount to composite fair value, placing it in the Fair Value range (±9.99%). The divergence between sources is very wide: GuruFocus (based on historical multiples) sees the stock as significantly undervalued, while pure DCF sources place it in a slight discount or fair value area. This dispersion is not a methodological error, but reflects genuine uncertainty over terminal growth rates.

> 📐 Discount 4.4% → Fair Value range → base score 5.50 | dispersion 76.2% MIXED → penalty −0.50 | post-penalty score 5.00 | Excellence Premium +1.38 (Business Score 9.38) → final score 6.38 (cap 6.50 not reached)

Score includes Excellence Premium +1.38 (Business Score 9.38/10) — cap 6.50 not applied.

B3.2 — Analyst consensus: 9.25

The sell-side consensus (source: TipRanks, 3-month window) is strongly bullish with 41 buy recommendations out of 43 analysts and an average target of $273.57, implying 36.8% upside versus the reference price. The distribution is polarized: almost all analysts still see significant appreciation potential, with targets ranging from $220 to $380 depending on different assumptions about the duration of the AI supercycle.

> 📐 Consensus (41/43 Buy, 95.3%) → Consensus_Score 9.49 | upside +36.8% → Upside_Score 9.00 (30–39.99% range) | U0 = 36.8% → w = 0.50 | B3.2 = 0.50 × 9.49 + 0.50 × 9.00 = 9.25

B3.3 — Relative valuation: 8.00

The AND condition is fully satisfied with relevant gaps on both dimensions. The current TTM P/E of 40.79x is significantly below the 5-year historical average of 69.67x (gap −41.4%) and below the peer group average of 62.20x (gap −34.4%). In absolute terms the multiple is not low, but relative to the company's history and reference industry, NVIDIA trades at a historically contained valuation — a configuration that reflects the speed of earnings growth more than a repricing of the business. The deep gaps on both dimensions support a score in the 8.00 range.

B3.4 — FCF & Net Shareholder Yield: 4.80

TTM FCF of $96.68 billion is exceptional in absolute terms, but relative to a Market Cap of $4.86 trillion it generates an FCF Yield of just 1.99%. The aggressive buyback (share reduction −1.17% YoY) contributes 0.83%, bringing Net Shareholder Yield to 2.84%, in the 2–4% range. The metric is structurally compressed by the market cap leverage effect, not by weakness in cash generation.

Combined profile: Solid business, positive outlook, attractive valuation.

Competitive Advantage and Moat

The competitive moat is Wide and expanding. The advantage is multidimensional: very high switching costs linked to the CUDA ecosystem (the de facto standard on which millions of developers' code is written), network effects linked to the installed base, economies of scale in R&D, and vertical control over the entire hardware-software stack. The more AI demand grows, the more the relative value of NVIDIA integration grows versus simple silicon. Competitors do not face a single barrier to overcome, but an entire ecosystem to rebuild.

General Cycle and Competitive Dynamics

The semiconductor cycle is expansionary and accelerated compared with the March analysis: Gartner has revised sector estimates upward to $1.30 trillion for 2026. Global hyperscalers are allocating capital in historic quantities — estimates indicate more than $630 billion in AI capex spending in 2026 — and NVIDIA is the primary recipient of these flows. AMD is advancing in inference, hyperscalers are developing custom silicon with increasing aggression, but the general-purpose AI training market remains structurally dominated by NVIDIA for at least the next 24–36 months.

Catalysts and Future Opportunities (Bull Case)

The full deployment of the Blackwell architecture and the Vera Rubin roadmap provide visibility on demand expansion over the next two technology cycles. The next earnings on 27/05/2026 will provide a Q1 FY2027 outlook expected at $78.00 billion (±2%). The growth of agentic AI opens a completely new addressable market that Morgan Stanley sees expanding chip demand well beyond GPUs. Sovereign computing by nation states creates non-cyclical and structurally stable demand. The remaining authorized buyback of $58.50 billion supports EPS in subsequent quarters.

Risks (Bear Case)

The priority risk is geopolitical: tightening export restrictions toward China (H20 ban) has already limited a historically relevant market, with a potential loss of more than $15 billion in revenue, and further escalation cannot be excluded. The second risk is structural: demand concentration among the main hyperscalers creates bilateral dependence — a slowdown in AI capex would be amplified into NVIDIA revenue. The third is technological: the emergence of customer custom silicon and alternative architectures could erode the marginal rent over the medium term, even without destabilizing the current position. The fourth is supply chain risk: total dependence on TSMC for physical production remains a structural non-diversifiable risk.

Excellent company near highs, FV justifies the price. GRADUAL ENTRY POSSIBLE.

Why it could be an opportunity

The intrinsic quality of the business is among the highest in the entire equity market (Business Score 9.38), with an expanding competitive moat and an earnings growth trajectory still supported by the product pipeline. The current P/E of 40.79x is historically low for NVIDIA and structurally below peers, meaning earnings growth has run faster than the price. Analyst consensus indicates 36.8% upside over 12 months. The weighted FV of $209.14 is above the current price, offering a modest margin of safety. Over a long horizon, the company remains one of the most robust compounders in the market.

Why it could be a risk

The stock trades near all-time highs (89.5% of the 52-week range), reducing the asymmetry of the risk/reward profile in the short term. Dispersion among DCF models is very high (76.2%), signaling genuine uncertainty over long-term growth estimates. Net Shareholder Yield of 2.84% is not particularly attractive for the capital committed at this valuation. Exogenous risks — export controls and dependence on TSMC — are structural, not temporary.

Price Target Table

This analysis is produced by the Score³ system for informational purposes only and does not constitute financial advice, a solicitation to invest, or a trading or investment recommendation. Data is collected from public sources and may contain errors or delays. Fair value estimates and price targets are model-based projections subject to significant uncertainty and do not represent certain forecasts. Investing involves risks, including the possible loss of invested capital. Always verify critical data against primary sources before making any investment decision. Past performance is not indicative of future results.