NVDA Up Just 3% While AMD Soars 171%: What Gives?
NVIDIA's revenue rose 85% yet its stock is up only 3.2% YTD while AMD surged 171%. Here's the divergence trade dominating semiconductor desks in July 2026.
Key Points
- NVIDIA's revenue grew 85% year-over-year to $81.6 billion, yet NVDA stock is up only 3.2% year-to-date while AMD has surged 171%, Micron 304%, and Intel 278% over the same period.
- The divergence reflects a market rotation toward AI infrastructure beneficiaries that were previously left behind, combined with a growing structural threat to NVIDIA's 81% AI chip market share from Broadcom's custom ASIC business — now guiding to $16 billion in Q3 AI revenue alone.
- The key watch event is NVIDIA's next earnings print, where investors will test whether the Vera Rubin platform's $7.8 million per rack pricing can sustain data center revenue growth rates above 90% year-over-year as hyperscalers simultaneously ramp competing custom silicon.
NVIDIA posted 85% revenue growth, $81.6 billion in annual sales, and 81% AI chip market share — and its stock is up just 3.2% in 2026. Meanwhile, AMD is up 171%, Micron has more than quadrupled with a 304% gain, and Intel has surged 278%. The VanEck Semiconductor ETF is up 59% year-to-date. The chip sector is in the middle of the most dramatic intra-sector divergence in at least a decade, and the rotation is accelerating even as semiconductor stocks broadly sold off on Wednesday on renewed trade and valuation concerns.
Why NVIDIA's Dominance Isn't Translating to Stock Performance
The paradox is frustrating for NVIDIA bulls, but the market's logic is internally consistent. NVIDIA's stock price going into 2026 had already embedded extraordinary growth expectations — the kind that 85% revenue growth struggles to beat when investors had priced in triple-digit acceleration. At a forward price-to-earnings ratio that has now compressed to less than 20x, NVIDIA is trading cheaper than the S&P 500's 21.5x forward multiple. That's a stunning fact: the company that controls the AI accelerator market, generates $75.2 billion in data center revenue annually — up 92% year-over-year — and whose H100 and B200 chips remain the de facto standard for large language model training, is now valued at a discount to the broader index.
The answer lies in the threat emerging from Broadcom's custom ASIC business. Broadcom guided Q3 AI semiconductor revenue to $16 billion — a figure that implies more than 200% year-over-year growth — against a confirmed customer list that reads like a who's-who of NVIDIA's largest accounts: Google, Meta, ByteDance, OpenAI, and one unnamed fifth hyperscaler. Broadcom's total custom chip backlog stands at $73 billion. Bloomberg Intelligence projects a 27% compound annual growth rate for custom ASICs through 2033, versus 16% for general-purpose AI accelerators like NVIDIA's GPUs. The market is beginning to price in structural share erosion at NVIDIA's core business, even as that business continues to grow in absolute terms.
There is also a ceiling-of-expectations problem specific to NVIDIA's next-generation Vera Rubin platform. Morgan Stanley Research published pricing estimates showing NVIDIA's VR200 NVL72 rack — the flagship Rubin-based system — will cost hyperscale cloud providers approximately $7.8 million per unit, nearly double the roughly $4 million price point for the prior GB300 generation. Memory alone now accounts for approximately $2 million of that rack cost, or roughly 25% of the total. At $55,000 per Rubin GPU for volume hyperscaler purchasers, NVIDIA's hardware remains premium-priced relative to the custom alternatives being built by Broadcom for the same customers. The Rubin platform's economics need to prove themselves in production deployments before investors will re-rate NVDA above its current compressed multiple.
The Rotation Beneficiaries and What's Driving Them
The stocks that have exploded in 2026 share a common characteristic: they were structurally underowned into the AI buildout cycle because investors assumed NVIDIA would capture nearly all the value. The re-pricing of that assumption has been violent. Intel's 278% year-to-date surge is the most striking case. Under CEO Lip-Bu Tan, Intel's IDM 2.0 strategy has moved from presentation slides to production reality: the 18A process node entered volume production in late 2025, and Microsoft and Qualcomm are confirmed foundry customers. NVIDIA itself injected credibility into the Intel recovery story by taking a $5 billion equity stake — a move that reads as both a supply-chain hedge and a strategic vote of confidence in 18A's viability for future chip generations.
Intel's April 2026 reacquisition of Apollo Fab 34 for $14.2 billion — funded through $7.7 billion in cash and $6.5 billion in debt — signals the company is not hedging its foundry ambitions. Q2 2026 guidance of $13.8 billion to $14.8 billion in revenue, with a non-GAAP gross margin of 39.0% and EPS of $0.20, represents a business still in transition but moving faster than Wall Street expected eighteen months ago. Micron's 304% gain is a different story: it reflects a deliberate strategic pivot away from the collapsing consumer smartphone market — IDC is forecasting a 13% year-over-year volume decline in global smartphones in 2026, the worst on record — toward high-bandwidth memory for AI data centers, where pricing power and margins are substantially higher. AMD's 171% surge reflects both its competitive GPU roadmap and the market's recognition that being the credible number-two in AI accelerators — behind a company growing at 85% — is still an extraordinary position to hold.
Chip Sector Sell-Off and What Comes Next
Wednesday's broad-based semiconductor sell-off hit every major name — NVDA, AMD, Broadcom, TSMC, Intel, Qualcomm, and Micron all traded lower — driven by a combination of trade uncertainty, valuation anxiety, and supply-chain noise. Reports of a potential Samsung strike and a reported stake sale by TSMC rattled investor confidence in the stability of global chip production. TSMC's own capacity constraints remain acute: CEO C.C. Wei publicly acknowledged at a shareholder meeting in early June that new U.S. production capacity is "far from enough," with additional sites planned in Japan and Germany. South Korea's government announced an $880 billion, 10-year investment plan covering semiconductors, AI infrastructure, and robotics — with Samsung and SK Hynix alone committing $518 billion toward new fabrication — but that capacity does not come online overnight.
The macro overlay compounds the pressure. At a 4.48% ten-year Treasury yield and CPI running at 4.2% year-over-year, the rate environment is not friendly to high-multiple growth names. The fact that NVDA has de-rated to below 20x forward earnings despite 85% revenue growth is partly a product of that rate pressure — and any uptick in yields from here would disproportionately weigh on the higher-multiple beneficiaries of the rotation, particularly AMD and Micron, which have seen their multiples stretch significantly as their stocks have rallied. TSMC's insider buying signal — Senior VP Choh Fei Yeap purchasing 1,000 shares for $55,780 and VP Shyue-Shyh Lin buying 3,000 shares for approximately $164,160 — offers a contrarian read that the sell-off may be creating opportunity in the foundry leader, which trades at approximately 16x 2026 earnings against a projected 30% revenue growth rate.
For traders navigating the divergence, the critical near-term events are the hyperscaler earnings in the final week of July. Microsoft reports July 29, with Azure's growth rate — 31% constant currency last quarter — the key number. Meta and Alphabet follow shortly after. Goldman Sachs projects four hyperscalers will collectively spend $725 billion on AI infrastructure in 2026, a 77% increase year-over-year. If those earnings confirm that capital expenditure commitments are being maintained or accelerated — and that the buildout is generating revenue, not just burning cash — NVIDIA's compressed multiple becomes the most obvious value trade in the sector. If hyperscaler management teams signal any moderation in AI capex in a 4.2% inflation environment, the entire chip sector faces a reset, and the rotation beneficiaries with stretched multiples will feel it first. The $255 level on SMH is the technical line to watch; a close below it on elevated volume would confirm the sell-off has moved from rotation to risk-off.
The Weekly Investor
Daily market analysis for active traders. Free.
