Nvidia (NVDA) vs S&P 500

Nvidia became the world's most valuable company in 2025 and held the position into April 2026. The $5.06 trillion market cap puts NVDA ahead of Microsoft ($3.5 trillion), Apple ($3.3 trillion), and Alphabet ($2.7 trillion). The company's share of the S&P 500 has grown from less than 1 percent in 2019 to approximately 7.5 to 8 percent in April 2026, the highest single-stock concentration in S&P 500 history.

The concentration matters mechanically for index investors. Every 1 percent move in NVDA corresponds to approximately 0.075 percent move in the S&P 500 from index weighting alone. Since November 2022, NVDA has risen 540 percent (from $14 split-adjusted to over $200) while the S&P 500 has roughly doubled. NVDA's contribution to S&P 500 returns over the cycle has been approximately 25 to 30 percent of total return, an unprecedented single-stock impact in modern index history.

NVDA's fiscal 2026 revenue of $215.9 billion (up 65 percent YoY) is dominated by data center revenue. Q4 fiscal 2026 data center revenue alone was $62.3 billion (up 75 percent YoY), implying annualized data center revenue of approximately $250 billion. The data center business is selling AI accelerator GPUs to hyperscalers (Microsoft, Google, Meta, Amazon, Oracle), AI-focused enterprises, and sovereign AI initiatives.

Hyperscaler capex commitments total approximately $500 billion annually in 2026, with NVDA capturing roughly half through GPU and accelerator sales. The capex cycle has produced unprecedented gross margins (over 75 percent gross, over 60 percent operating) that exceed any prior tech hardware leader in history. The concentration of S&P 500 returns in NVDA reflects markets pricing this AI infrastructure cycle as a generational capital reallocation event, similar to the 1990s internet buildout but with higher near-term cash flow conversion.

NVDA's 7.5 to 8 percent weight in the S&P 500 is the highest single-stock concentration in modern history. Prior peaks: AT&T at approximately 13 percent in 1932 (before the index reformed), General Motors at 7.5 percent in 1956, IBM at 6 percent in 1985, Microsoft at 4.9 percent in 1999, Apple at 7.1 percent in 2024 (briefly before NVDA overtook).

The concentration creates two-way risks. On the upside, continued NVDA outperformance lifts the S&P 500 disproportionately. On the downside, NVDA-specific weakness (regulatory action on China sales, AMD competitive pressure, hyperscaler capex slowdown) would drag the broader index more than fundamentals alone would warrant. Index investors holding SPY are effectively making a 7.5 percent bet on NVDA whether they intend to or not. For diversification purposes, equal-weight S&P 500 (RSP) reduces NVDA concentration to roughly 0.2 percent versus 7.5 percent in cap-weighted SPY.

From the ChatGPT release in November 2022 through the October 2025 NVDA peak, NVDA rose approximately 1,300 percent (from $14 split-adjusted to $190). SPY rose approximately 75 percent over the same window. The NVDA/SPY ratio expanded from 0.034 in November 2022 to 0.27 in October 2025, an approximately 7-fold increase in relative performance.

The outperformance was nearly continuous: NVDA outperformed SPY in 8 of 11 quarterly windows from Q4 2022 through Q3 2025. The brief underperformance windows (Q1 2024, Q3 2024, Q1 2025) corresponded to specific events: Blackwell architecture delay concerns, August 2024 yen carry unwind, and DeepSeek China LLM concerns. Each underperformance window resolved within 4 to 8 weeks as the AI capex narrative reasserted itself. The pattern produced one of the cleanest single-stock leadership trends in modern history.

NVDA at $208 in April 2026 is below the late-2025 peak near $230 but well above the trailing-12-month average. Q4 fiscal 2026 results released February 25, 2026 produced minor disappointment relative to the highest expectations: revenue of $68.1 billion versus consensus $69.5 billion, with forward guidance suggesting some moderation in growth pace.

SPY at $708 has held within 1 percent of its all-time high of $712 through April. NVDA has therefore underperformed SPY year-to-date 2026 by approximately 5 to 7 percentage points, the largest sustained NVDA underperformance since 2018. The configuration suggests AI capex narrative has matured: hyperscaler spending continues but at decelerating pace, and NVDA forward growth expectations are modestly compressing. A clean Q1 fiscal 2027 result (released late May 2026) could re-energize the narrative; disappointment could extend the underperformance.

Three factors drive NVDA-specific moves disconnected from SPY. First, NVDA earnings cycles: each quarterly release moves NVDA 5 to 15 percent typically, with limited SPY response. Second, AI capex commentary: hyperscaler quarterly capex updates (Microsoft, Google, Meta, Amazon, Oracle) move NVDA more than SPY because the SPY exposure to AI capex is diluted across many holdings while NVDA is the pure play.

Third, regulatory and competitive shocks: US export restrictions on China sales (Q4 2023, Q3 2024), AMD MI300 competitive announcements, custom hyperscaler silicon (Google TPU, Amazon Trainium, Microsoft Maia) all produce NVDA-specific moves. The April 2026 environment has been relatively quiet on these fronts. The next major catalyst is the May 2026 Q1 fiscal 2027 earnings release, which markets expect to reveal whether the AI capex deceleration is structural or transitory.

The Iran war that began February 2026 has produced minimal direct NVDA impact. NVDA's revenue base is dominated by US, China, and Western European hyperscalers; Iran-related geopolitical risk does not flow through to GPU demand. NVDA stock has traded sideways through the conflict, in line with broader tech sector behavior, while energy stocks (XLE) have rallied substantially.

The indirect effect on NVDA has been through broader risk-off flows. NVDA has higher beta than SPY (~1.4 to 1.6), so the modest SPY pullback during the conflict's peak (March 27, 2026 with VIX at 31) produced larger NVDA drawdowns. NVDA recovered quickly with the broader market. The Iran war effect on NVDA has been less than the AI capex narrative pressure, which is the more important driver in April 2026.

Approximately 50 percent of NVDA's data center revenue comes from the five largest hyperscalers (Microsoft Azure, Google Cloud, Meta, Amazon AWS, Oracle Cloud). The customer concentration creates two distinct risks. First, hyperscaler capex deceleration: any of the five reducing AI capex commitments would compress NVDA's revenue trajectory. Second, internal alternatives: each hyperscaler has been developing custom AI silicon (Google TPU, Microsoft Maia, AWS Trainium, Meta MTIA) to reduce NVDA dependence.

Through 2024 to 2025, custom silicon represented less than 15 percent of hyperscaler AI compute. Markets have generally accepted that NVDA will retain dominant share through at least 2028, but the eventual shift to internal alternatives is a structural risk. Q1 to Q3 2026 hyperscaler capex commentary will be the first window where markets can assess whether the customer concentration risk is materializing or whether NVDA continues to capture incremental capex growth.

NVDA represents the cleanest direct AI capex play, but the broader AI trade extends across multiple sectors. Power utilities (NextEra, Constellation, Vistra) benefit from data center electricity demand. Real estate (Equinix, Digital Realty) benefits from data center construction. Networking (Arista, Cisco) benefits from data center buildouts. Memory (Micron, SK Hynix) benefits from HBM demand for AI training.

For investors holding SPY, the AI exposure is diluted: NVDA at 7.5 percent weight, plus Microsoft (5.7 percent), Alphabet (combined 6.7 percent), Amazon (5.5 percent), Meta (3.7 percent), Apple (7.6 percent) totals roughly 36 percent of SPY in AI-exposed names. Long NVDA versus short SPY isolates the pure NVDA-specific component of the AI trade, separating it from the broader Magnificent 7 dynamics.

For practical use: track the NVDA/SPY ratio. April 2026 ratio is approximately $208 / $708 = 0.294. The ratio peaked at approximately 0.32 in October 2025 and bottomed at 0.034 in November 2022 (pre-ChatGPT). Historical context: the ratio of any single-stock peak versus index has typically been short-lived, with mean reversion within 1 to 3 years.

For pair trading: long NVDA / short SPY captures pure AI capex exposure with hedged broad market beta. Short NVDA / long SPY is a defensive bet that benefits if AI capex disappoints. Position sizing should account for NVDA's higher volatility (typical 30-40 percent annualized vs SPY 15-20 percent). The April 2026 NVDA/SPY ratio at 0.294 is below the October 2025 peak of 0.32 but well above the long-run average. The current configuration favors slight NVDA underperformance until the Q1 fiscal 2027 earnings release in late May 2026 clarifies the AI capex trajectory.