In the second quarter of 2026, which just concluded for US stocks, the artificial intelligence (AI) investment wave witnessed a spectacular shift in market leadership.
The long-dominant supergiant of AI computing power, NVIDIA (NVDA), significantly slowed its blistering pace of gains during the quarter.
As investors shifted their focus from the "pickaxe sellers" of AI computing like NVIDIA to the broader AI infrastructure supply chain, shares of Micron Technology (MU), Intel (INTC), and Advanced Micro Devices (AMD) collectively soared, adding a combined market value of approximately $2 trillion.
This rally, led by the "AI enablers," propelled the Philadelphia Semiconductor Index (SOX) to surge 81% in Q2, marking its strongest quarterly performance on record.
In contrast, the AI chip leader NVIDIA (NVDA) gained only 15% for the quarter, becoming one of the weakest performers in the SOX index.
This extreme divergence signals a deepening spread of AI investment logic from "computing power monopoly" to "empowerment across the entire industry chain."
Capital Rotation: From Cloud Giants to Enabling Chipmakers
The core driver of this market movement is a systematic rotation of capital from AI hyperscale cloud providers to the AI infrastructure supply chain.
Barclays analyst Anshul Gupta explicitly noted in a Tuesday report that funds are shifting from AI hyperscalers to AI-enabling chip companies, with the market enthusiasm subsequently transmitting to the semiconductor sector, driving a massive rally.
Performance among hyperscale cloud providers was mixed this quarter: Alphabet (GOOGL) led with a 24% gain, Amazon (AMZN) and Microsoft (MSFT) posted moderate increases, while Meta Platforms (META) fell nearly 2%, ranking at the bottom.
Cantor Fitzgerald Senior Managing Director and Technology Analyst CJ Muse remarked that the past six months have been entirely about the market betting heavily on AI infrastructure, but now people are questioning its sustainability and whether there should be concern.
This quarter witnessed a profound structural shift in the flow of global capital.
Investors are rapidly broadening their AI portfolios, seeking high-growth targets beyond NVIDIA.
The three chip companies—Micron, Intel, and AMD—added a staggering combined market cap of around $2 trillion in Q2, a figure that not only reshaped Wall Street's landscape but also propelled these three firms into the 10th, 11th, and 12th spots in the ranking of US tech company market values.
Meanwhile, the usual market bellwether, NVIDIA (NVDA), posted a "mere" 15% gain in Q2; although its revenue growth remained exceptionally strong, its stock price increase was significantly surpassed by its peers.
Memory Chips: AI Computing's "New Bottleneck," Micron Surpasses $1 Trillion
Memory chips were the most explosive sector in this rally.
As AI workloads drive exponential growth in demand for high-bandwidth memory (HBM), memory chips have become the "new bottleneck" in the semiconductor industry.
Shares of Micron Technology (MU) surged over 240% in Q2, adding approximately $920 billion in market value.
The company's latest quarterly revenue grew over fourfold year-over-year, with gross margins jumping to 84.9% from 39% a year ago.
Micron stock is up 270% year-to-date and a staggering 860% over the past 12 months.
In May, its market capitalization broke through the $1 trillion mark.
CEO Sanjay Mehrotra earlier this year discontinued the company's 29-year-old consumer business, shifting capacity entirely to data centers and moving from spot chip sales to long-term contracts to maintain high prices.
Currently, Micron can only meet 50% to 75% of customer demand and is deliberately controlling the pace of new fab construction to ensure supply does not outstrip demand.
The collective surge in the memory sector is not limited to Micron.
SanDisk (SNDK) has soared 764% year-to-date, making it the best-performing stock in the S&P 500; Western Digital (WDC) and Seagate Technology (STX) also rank among the top gainers.
The three memory giants—Micron, SK Hynix, and Samsung—all now boast market caps exceeding $1 trillion.
CLSA analyst Bhavtosh Vajpayee noted in a Monday report that HBM content in AI chips will continue to increase, expected to exceed 500GB by 2030, with supply-demand balance for memory components unlikely to be restored before then.
He wrote that it is premature to fear a downturn, adding that while there will be down cycles in AI, the oversupply scenario the market worries about is not expected until 2030 or the second half of 2029 at the earliest.
CPU Renaissance: The "Second Spring" for Intel and AMD
AI's accelerating penetration from the cloud to edge devices is bringing an unexpected "second spring" for traditional CPU manufacturers.
Shares of Intel (INTC) surged 216% in Q2, adding about $480 billion in market value.
The company is building chip plants in the US and benefiting from a strong recovery in CPU demand as AI shifts toward devices.
Intel previously forecast Q2 revenue between $13.8 billion and $14.8 billion, far exceeding the analyst consensus of $13 billion.
Shares of AMD (AMD) nearly tripled in Q2, adding approximately $615 billion in market value.
In May, AMD's earnings report showed Q1 revenue beat expectations, and it provided strong guidance for Q2 data center business, pushing its stock price to a record high above $430.
CEO Lisa Su stated that customer engagement with the new MI450 series AI accelerators and Helios rack systems is "increasing," with forecasts from lead customers exceeding the company's initial expectations.
AMD also produces GPUs but remains far behind NVIDIA in that market.
Notably, Intel's forward price-to-earnings ratio is around 100x, while AMD's valuation is also elevated, reflecting extremely high market expectations for both companies' transformation prospects.
Full AI Infrastructure Chain Surge: Marvell and Arm Lead Gains
Beyond memory and processors, other segments of the AI infrastructure supply chain also experienced massive stock price increases.
Networking equipment maker Marvell Technology (MRVL) saw its shares rise about 200% in Q2.
ARM Holdings (ARM), which provides technology and architecture licenses to other chip firms, gained 134% over the same period.
However, Arm's forward P/E ratio has exceeded 140x, placing it in severely overvalued territory by traditional metrics.
Sector Valuation and Risk: Concerns Beneath the Feast
Despite the fiery rally, the semiconductor sector's valuation and volatility have entered extreme territory.
Valuations are elevated.
The SOX index currently trades at a forward P/E of about 26x, significantly above its 10-year average of 19x.
Intel trades at a forward P/E of around 100x, and Arm over 140x.
In contrast, NVIDIA's forward P/E is just 18x, its lowest since 2018.
The warning bells of a valuation bubble and the reality of severe volatility remind every momentum investor that historic rallies often come with historic risks.
Volatility has surged.
The Cboe Semiconductor ETF Volatility Index has risen 83% this year, on track for its largest annual increase on record.
This month, the SOX index had only one trading day with a closing range below 1%, having once gained 7.9% in a single day only to fall over 10% later.
Fund flows have shown reversals.
The VanEck Semiconductor ETF (SMH) plunged 7.3% last week, its worst weekly performance since April 2025, before rebounding quickly.
Mizuho TMT specialist Jordan Klein believes the chip stock decline and software stock rebound were "more about quant funds and relative return managers doing quarter-end rebalancing than any concerning or ominous signal."
Retail enthusiasm remains strong.
Vanda Research data shows Micron was the most-bought stock by retail investors on Monday, with inflows of about $30.8 million; retail inflows over the past two trading days exceeded $103 million, the highest since mid-March.
Nevertheless, Nomura analyst Aaron Jeng stated that a pullback after such a rapid price surge is healthy, especially when seeing risks that need to be digested—such as potentially the largest-ever component supply mismatch and the sustainability of hyperscaler capital expenditures.
He added that the peak of this cycle has likely not yet been reached.
The New Paradigm of "AI Power Shift": AI Investment Enters 2.0 Era
The second quarter of 2026 saw the semiconductor industry undergo a profound paradigm shift.
The market has recognized that AI is not a single-company story but an ecosystem requiring collaboration across the entire industry chain—from HBM memory to CPU compute, from networking equipment to chip architecture licensing, each link is creating value.
Analysts have described the quarter's market action as a "power shift" within the AI industry: investors positioned themselves in companies producing chips that complement NVIDIA while betting that massive capital expenditure expansion in AI data centers would boost earnings for more companies across the supply chain.
From "only NVIDIA" to "the entire semiconductor supply chain," from "AI chips" to "memory, networking, foundry, IP," the reach of capital is extending at an unprecedented pace.
The VanEck Semiconductor ETF's quarterly gain of 71% marks its best performance in its 26-year history, highlighting the massive scale of capital flowing into the sector.
Thornburg Investment Management portfolio manager Sean Sun noted that investors are chasing bottlenecks in the semiconductor space, which currently favor memory chips and also the recovery of Intel's foundry business.
The US semiconductor storm of Q2 proved one thing to all: the AI revolution has evolved from the initial "single-core drive by NVIDIA" to a comprehensive resonance across the entire data center supply chain, edge-side hardware, and network connectivity infrastructure.
Investors are beginning to realize that a trillion-dollar-scale AI data center not only needs NVIDIA's GPUs but also Micron's high-bandwidth memory, Intel and AMD's CPUs for scheduling, and Marvell Technology's high-speed network transmission.
Looking ahead, although short-term valuations for companies like Micron and Intel may experience volatility after their tripling surges, there are no signs of global tech giants scaling back capital expenditure on AI computing infrastructure.
This comeback drama directed by "non-NVIDIA" chip stocks not only shows the market the broad-based benefits of the AI dividend but also suggests that internal sector divergence and volatility may have only just begun.
Comments