NVIDIA: Growth, Durability, and Mispricing in the AI Infrastructure Buildout

• NVIDIA is being valued as though its growth is cyclical, despite strong evidence that AI-driven demand is structural and still in its early stages.

• Forward valuation has compressed meaningfully due to earnings expansion, not weakening fundamentals, leaving the company trading at levels inconsistent with its growth profile.

• Even under conservative assumptions, such as multiple contraction and earnings underperformance, downside appears limited relative to the potential upside.

Overview

NVIDIA has solidified its position as the central enabler of the modern AI ecosystem, with its data center segment now accounting for approximately 90% of its total revenue. It maintains a dominant position in data center GPU markets as a whole, with an estimated share approaching 90% of the TAM, driven by performance leadership, ecosystem integration, and sustained technological advancement.

Despite this, recent sentiment appears to have shifted toward concerns surrounding the durability of AI-driven demand. Recent discourse has increasingly focused on the possibility of an ‘AI bubble,’ with some drawing comparisons to the dot-com era and the 2008 financial crisis. Questions regarding hyperscaler spending, infrastructure constraints, and potential overcapacity have contributed to an overly cautious market view.

While these concerns are not without merit, they appear overstated relative to the scale and trajectory of underlying demand. NVIDIA's current price is now at a point that is only sensible if investors either have an overwhelmingly pessimistic projection of earnings growth compared to consensus estimates or an extreme degree of apprehensiveness about NVIDIA's deserved earnings multiple.

The Engine: AI Demand & Data Center Expansion

At the core of NVIDIA’s growth is the rapid expansion of AI infrastructure. Industry data suggests that data center spending is expected to grow at a CAGR of approximately 15%, with GPU-related demand expanding significantly faster, in some estimates exceeding 30% annually.

This divergence is important. GPUs are not simply a component of the data center, but increasingly the bottleneck resource. As AI workloads scale in complexity and volume, compute requirements rise disproportionately, reinforcing demand for high-performance accelerators.

Recent commentary from management, including projections discussed at NVIDIA’s GTC conference, forecasts a trillion-dollar revenue opportunity tied to AI infrastructure—specifically data center and GPU demand—through fiscal 2027. While the scale of these projections is significant, they remain directionally consistent with broader industry forecasts and capital allocation trends.

The Moat: Market Position & Pricing Power

NVIDIA’s competitive advantage is not solely a function of its hardware performance, but of its ecosystem depth. Its GPU architecture, coupled with a deeply entrenched software stack, has created a level of integration that is increasingly difficult to replicate at scale.

This has translated directly into pricing power. With demand for its latest Blackwell architecture significantly outpacing supply, NVIDIA has been able to sustain operating margins in excess of 50% since 2024. At present, Blackwell-based systems account for a substantial portion of data center revenue, reinforcing both the company’s technological leadership and, more importantly, its ability to monetize it effectively.

Short-Term Dynamics: Constraints vs Demand

A central concern in the current market narrative is that AI infrastructure investment has outpaced real demand, particularly among hyperscale cloud providers. This argument is often framed around the idea that GPU supply is exceeding the capacity of data centers and power infrastructure to support it. While these constraints are real, the conclusion that they imply weakening demand is less convincing.

Hyperscalers have continued to signal aggressive capital expenditure plans extending into full-year 2026 and beyond. Independent research, including estimates from McKinsey, suggests that global investment in AI-related data center infrastructure could reach several trillion dollars over time, underscoring the scale of the opportunity.

What is currently being interpreted as potential overcapacity is more accurately a mismatch between compute availability and supporting infrastructure. In practice, this represents deferred deployment rather than diminished demand, an important distinction, and one evidenced by hyperscaler CAPEX commitments. As a result, these constraints are more likely to impact the timing of growth rather than its magnitude.

The Valuation & Margin of Safety

The current valuation of NVIDIA reflects a meaningful degree of skepticism regarding the durability of its growth. Specifically, the market appears to be discounting the possibility that recent earnings expansion represents a temporary surge, with AI spending increasingly treated as cyclical rather than infrastructural.

This view appears inconsistent with even the most conservative Wall Street estimates for NVIDIA’s revenue and earnings in the coming years, as well as with the scale of AI-related capital investment already outlined by hyperscalers. NVIDIA has also consistently outperformed consensus estimates, often by notable margins. Over the last twelve quarters, NVIDIA has exceeded quarterly EPS estimates by an average of approximately 11%, and roughly 6% over the most recent five quarters

At current levels, NVIDIA trades at approximately 22x forward earnings based on consensus estimates of roughly $8.25 in EPS. Adjusting for the company’s consistent history of earnings outperformance (even assuming the more recent trends of ~6% to be conservative), this implies an effective forward multiple that is really closer to 20–21x.

This is particularly notable given the company’s growth profile. Consensus estimates project earnings growth of approximately 50–60% year-over-year, placing NVIDIA’s PEG ratio well below 1.0 and near multi-year lows.

The relationship between earnings growth and valuation becomes even more apparent when viewed over time. While NVIDIA’s share price has appreciated significantly, its earnings have grown at an even faster rate, resulting in meaningful multiple compression. In effect, NVIDIA’s growth is being priced more cheaply today than at any point in recent years.

To better frame the opportunity created by this valuation, we can analyze a set of outcomes that incorporates a range of valuation scenarios and varying assumptions for earnings growth and multiple stability or contraction.

It is reasonable to assume NVIDIA’s trailing P/E could contract to approximately 30x, which would represent a roughly 20% decline from its current PE. Additionally, in four of the five scenarios presented, earnings are assumed to fall short of consensus by 12% or more, a very conservative outlook considering NVIDIA's history of beating estimates.

Still, even under these assumptions, where earnings growth falls short of expectations and valuation multiples contract, the implied downside remains relatively contained, even in worst case scenarios (see 'Full Bear' case below). In contrast, even neutral scenarios suggest meaningful upside.

Given NVIDIA’s market position, earnings projections, and structural growth drivers that for the moment quell the largest risk concerns, a forward multiple in the range of 25–30x appears reasonable, if not conservative. At those levels, the company’s current valuation implies a clear disconnect between price and underlying earnings power.

Risks

Despite the strength of the thesis, several risks warrant consideration:

• Export restrictions and geopolitical exposure: Limitations on chip sales to China could impact a historically meaningful portion of data center revenue. Broader geopolitical tensions, including those involving China and Taiwan, remain a longer-term consideration, though their timing and likelihood remain uncertain.

• Hyperscaler concentration: A significant share of revenue is tied to a small number of large customers, increasing exposure to shifts in their capital allocation strategies.

• Infrastructure bottlenecks: Power and data center constraints may delay deployment timelines, even if underlying demand remains intact.

While these risks are real, they are more likely to influence the timing of growth rather than its overall trajectory. Many of these concerns are longer-term in nature and appear less relevant to the current dislocation between NVIDIA’s valuation and its earnings outlook.

The current valuation disconnect is not a function of these risks being immaterial, but rather of their timing. NVIDIA’s pricing implies a level of near-term disruption that simply isn't supported by current demand trends, capital expenditure commitments, or earnings outlook. Even in scenarios where these risks do begin to materialize sooner rather than later, the inherent margin of safety at NVIDIA's current price leaves room for meaningful deterioration in operating performance before its valuation poses serious downside risk.

Conclusion

NVIDIA occupies a uniquely advantaged position at the center of one of the largest technological buildouts in modern history. Its combination of market share, pricing power, and earnings growth is rare, particularly at its scale. Yet despite these attributes, the company is currently valued at levels that suggest a level of skepticism that appears inconsistent with the durability of its growth. This creates an opportunity.

The market is effectively pricing NVIDIA as though it is nearing a cyclical peak already, while the underlying data suggests it is closer to the middle, or even the early stages, of a structural expansion. In this context, the investment case is less about predicting continued success and more about recognizing that much of that success is not fully reflected in the current valuation.

---

Sources & References

KPMG, Investing in GPUs: Market Growth Outlook

McKinsey & Company, The Cost of Compute: A $7 Trillion Race to Scale Data Centers

CNBC, NVIDIA GTC 2026: Jensen Huang Keynote and Forward Outlook

CNBC, NVIDIA China Chip Sales and Export Controls

Axios, NVIDIA and China Market Developments

IoT Analytics, Data Center GPU Market Share Estimates

This article reflects the views and analysis of The Benjamin Fund as of the date of publication. Information presented is derived from company disclosures, industry reports, and third-party research believed to be reliable, but accuracy and completeness are not guaranteed. This material is for informational purposes only and should not be construed as investment advice or a recommendation to buy or sell any security.