Nvidia, the Infinite Premium the Market Has Stopped Paying
Equity Research · Valuation
Nvidia continues to grow at an exceptional pace, yet the market has begun to distinguish between the quality of technological leadership and the price it is willing to pay for it. The correction from the May peak is the measure of that distinction.
1. Executive summary
Nvidia burned through nearly $800 billion in market capitalisation at the most acute phase of the correction that began in mid-May. Even for a company accustomed to moving entire indices on its own, the figure remains staggering. Market capitalisation has, however, recovered towards $4,940 billion, reaffirming the group as the most valuable company in the US market.
The essential distinction lies between two different metrics. Market capitalisation measures the size of the company; the valuation multiple measures the price the market is willing to pay for expected future earnings. Nvidia retains an exceptional scale, while the narrative premium ascribed to its AI leadership is compressing.
The thesis of this analysis is that the market is separating industrial quality from the price paid for that quality. Fundamentals remain very strong, while the forward P/E has fallen to historically low levels for the stock. The focus shifts from revenue growth to its sustainability, and to the discipline with which hyperscalers convert capital expenditure into returns on capital.
2. Reading the signals
A broad but orderly correction. From the mid-May peak, around the $235 level, the stock retreated towards $204, with market capitalisation falling far enough to erase nearly $800 billion before recovering part of that ground.
Multiple compression. The forward P/E has declined to approximately 21.7 times expected earnings, against a five-year average of 72 times. This is the clearest signal: the market is paying less for the same growth.
Relative underperformance. Year-to-date, the stock has gained approximately 5%, while the SOXX semiconductor index has risen nearly 60% and AMD has more than doubled. Leadership remains solid, but the premium attached to it is narrowing.
Fundamentals intact. Revenue +85%, Data Centre +92%, gross margin at 74.9%, and free cash flow of $48.6 billion in a single quarter. The numbers alone do not explain the correction.
3. The number that changes the narrative
The drawdown from the May peak came close to $750–800 billion at its trough — a figure exceeding the market capitalisation of many large-cap global companies. The data should nonetheless be read in context: over the trailing twelve months the stock remains up nearly 29%, and its market capitalisation, close to $4,940 billion, confirms it as the most valuable company in the US market.
In the chart we observe the multi-year trajectory of market capitalisation: the near-vertical ascent after 2023 makes clear why the recent correction, enormous in absolute terms, remains contained relative to the underlying trend.
4. Price, volatility and technical demand
On the technical front, the stock reached a high of around $235 in mid-May before pulling back towards the $200–205 level. Despite the correction, it retains a gain of nearly 29% over the past twelve months: the downward phase represents a normalisation within a still-positive trend, not a confirmed reversal.
Relative underperformance versus the sector is the true signal of the extended session. With the stock up approximately 5% year-to-date against a SOXX approaching +60%, the market is rotating towards other beneficiaries within the AI supply chain, pricing the GPU leader with greater discipline.
In the chart we observe the stock's path over the past twelve months: the spring peak and the subsequent decline, within an annual trend that remains positive.
5. The platform the market is pricing
Nvidia's valuation incorporates far more than a single product cycle. The market is pricing an accelerated-computing platform — comprising GPUs, the CUDA software ecosystem, networking and rack-scale systems — positioned at the heart of artificial intelligence infrastructure.
The pool of expenditure feeding this platform remains vast. Market estimates point to hyperscaler capex in the range of $650–725 billion in 2026, with projections exceeding $1,000 billion in 2027. As long as that spending grows, Nvidia's revenue engine remains on; the question is for how long, and with what return for those sustaining it.
6. Data Centre: the group's engine
The Data Centre segment is now Nvidia's economic core. In the first quarter of fiscal year 2027, it generated $75.2 billion in revenue, up 92% year-on-year and accounting for more than 90% of total revenues. This is where demand for training and inference accelerators is concentrated.
This concentration is simultaneously a strength and a vulnerability. It makes the group the default choice for those building AI infrastructure, but exposes revenues to the spending pace of a limited number of large customers and to their progressive vertical integration.
7. Fundamentals and profitability
The first-quarter results for fiscal year 2027 remain those of a company with exceptional fundamentals. The table summarises the key line items, together with the relevant context.
| Item (Q1 FY2027) | Value | Reading |
|---|---|---|
| Revenue | $81.6bn (+85%) | Third consecutive quarter of acceleration. |
| Data Centre | $75.2bn (+92%) | Over 90% of revenues: the engine and the concentration risk. |
| GAAP gross margin | 74,9% | Profitability consistent with infrastructure leadership. |
| Operating income | $53.5bn | Operating efficiency at increasing scale. |
| Non-GAAP net income | $45.5bn | Non-GAAP EPS of $1.87. |
| Free cash flow | $48.6bn | Cash generation funding buybacks and the dividend. |
| Capital returns | +$80bn buyback | New authorisation; dividend raised to $0.25. |
| Q2 FY2027 guidance | $91.0bn (±2%) | +11% quarter-on-quarter: growth still in progress. |
In the chart we observe the trajectory of annual revenues: growth, nearly flat for years, accelerates sharply after 2023, driven by the Data Centre segment.
The compression of the forward P/E coexists with these figures: the market is recalibrating the valuation premium applied to growth that remains exceptional.
8. The Supply Chain: HBM and Memory
The AI supply-chain bottleneck is shifting across multiple links in the chain. HBM (High Bandwidth Memory), produced by SK Hynix, Micron and Samsung, has become a critical resource for accelerators. Networking, storage and physical infrastructure are also taking on greater significance.
The chart shows memory's share of hyperscalers' total capex: the estimated weight rises from 7.8% in 2023 to 30.0% in 2026 and 36.2% in 2027. Value is redistributing along the supply chain, keeping Nvidia at the centre while broadening the universe of beneficiaries.
9. Who Is Funding AI Capex
The most sensitive issue concerns the hyperscalers — the major cloud operators and platforms financing AI infrastructure: Microsoft, Alphabet, Amazon, Meta and others. Estimates point to aggregate capex of approximately $650–725 billion in 2026, with projections exceeding $1 trillion in 2027.
The risk for Nvidia hinges on capital discipline. Operators are asking when this spending will generate visible revenues, margins and free cash flow for those bearing the cost. The gap between AI expenditure and the revenues it generates is widening, and the market has begun to price this in: Nvidia's revenues depend on infrastructure appetite, but the sustainability of that appetite is being judged on customers' balance sheets.
10. Custom Chips and Competition
Alongside capex pressure comes the issue of proprietary chips. Google is developing TPUs, Amazon is working on Trainium and Inferentia, OpenAI is collaborating with Broadcom on custom inference solutions, while DeepSeek is reportedly developing its own chip to reduce dependence on Nvidia and Huawei.
Inference — the execution of already-trained models — could become the most fertile ground for specialised, lower-cost chips. Full substitution of Nvidia remains a remote scenario, while the market may begin to price in a normalisation of future pricing power across certain portions of the workload. The conditional tense remains necessary: the industrial direction is clear; the timing is not.
11. Valuation: The Premium Compresses
At the current price, the stock trades at approximately 21.7 times forward earnings — a level close to the S&P 500 index average and well below Nvidia's five-year average of approximately 72 times. Goldman Sachs considers this valuation "compelling", arguing that it does not fully reflect the expected revenue growth tied to the acceleration in AI capex.
Analyst consensus remains constructive: across 61 ratings, the prevailing view is a Strong Buy, with a twelve-month average price target of $301.62, a high of $500 and a low of $180. The dispersion of targets signals that the debate centres on growth expectations, not on business quality.
The two charts set alongside each other the multiple compression and the consensus snapshot: the historical premium deflates, while analysts maintain targets largely above the current price.
| Valuation Metric | Level | Reading |
|---|---|---|
| Current Forward P/E | ~21.7x | Close to market average: compressed premium. |
| 5-Year Average P/E | ~72x | Reference point from the peak of AI enthusiasm. |
| Analyst Consensus Target | 301,62 $ | Approximately +50% from the current price (61 analysts). |
| Target Range | 180 - 500 $ | Wide dispersion on growth assumptions. |
12. Investment View
The area-by-area reading combines industrial quality with valuation risk. Judgements are probabilistic and do not constitute investment recommendations.
| Area | View | Rationale | What to Monitor |
|---|---|---|---|
| Fundamentals & Cash | Constructive | Exceptional growth, margins, and free cash flow. | Margin resilience during the scaling phase. |
| Data Centre | Constructive | +92% and over 90% of revenues. | Guidance and order momentum. |
| Valuation | Optionality | Compressed multiple versus history, but tied to growth. | Earnings revisions and forward P/E. |
| Hyperscaler Capex | Watch | Enormous spending but under scrutiny for sustainability. | Capex/revenue gap in client balance sheets. |
| Custom Chip Competition | Risk | TPUs, Trainium, custom solutions, and inference. | Adoption of proprietary chips in inference workloads. |
| Price Momentum | Neutral | Correction within a still-positive trend. | Sector rotation and resilience of the $200 level. |
13. Scenario Map
The scenarios outline the possible trajectories over twelve months. Probabilities are editorial estimates, useful for ranking risks, not point forecasts.
| Scenario | Indicative Probability | Driver | Implication |
|---|---|---|---|
| Base Case | ~55% | Solid Data Centre growth, capex still expanding, stable multiple. | Gradual re-rating towards consensus targets. |
| Upside | ~25% | Re-acceleration of capex towards $1 trillion and renewed margin expansion. | Return of the valuation premium, push towards the upper end of targets. |
| Downside | ~20% | Hyperscaler capital discipline, advance of custom chips, downward revisions. | Further multiple compression and price pressure. |
14. What to Monitor
Fundamentals & Valuation
- Data Centre guidance and order momentum.
- Gross and operating margin resilience.
- Forward earnings revisions and forward P/E.
- Capital returns: buybacks and dividend.
Supply Chain & Competition
- Hyperscaler capex and the capex/revenue gap.
- HBM memory capacity and orders.
- Custom chip adoption in inference workloads.
- Rotation towards other AI supply-chain beneficiaries.
15. Upcoming Events
| Event | Why It Matters | Positive Signal | Negative Signal |
|---|---|---|---|
| Q2 FY2027 Results | Verification of the $91 billion guidance. | Revenue and Data Centre above expectations. | Margin decline or cautious guidance. |
| Hyperscaler Earnings | Sustainability of AI capex. | Capex confirmed or revised upward. | Cautious tone on spending. |
| Custom Chip Updates | Pressure on pricing power. | Delays or limited adoption. | Large-scale deployment in inference. |
| HBM Memory Data | Supply-chain bottleneck. | Rising capacity and stable prices. | Shortage or rising costs. |
16. Operational Conclusion
Principal Sources Consulted
| # | Source | Use in the Analysis |
|---|---|---|
| 1 | Nvidia — Q1 FY2027 Results (20 May 2026) | Revenues, Data Centre, margins, EPS, FCF, buybacks, Q2 guidance. |
| 2 | StockAnalysis — NVDA (9 Jul 2026) | Price $204.12, +28.99% over 12 months. |
| 3 | CompaniesMarketCap — Nvidia (Jul 2026) | Market capitalisation ~$4,940bn and historical record 1999–2026. |
| 4 | StockAnalysis — NVDA Forecast | Consensus target $301.62, 61 analysts, Strong Buy. |
| 5 | TradingKey / Goldman Sachs (Jul 2026) | Forward P/E 21.7x vs. 5-year average 72x; capex $650bn → $1 trillion. |
| 6 | CNBC — Big Tech Capex to Exceed $1 Trillion by 2027 | Hyperscaler capex trajectory. |
| 7 | Tom's Hardware / SemiAnalysis (3 Apr 2026) | Memory share of hyperscaler capex. |
| 8 | Times of India / Reuters (7 Jul 2026) | Report on DeepSeek inference chip (unconfirmed). |