AI Capex Anxiety Splits Wall Street: Intel's Record Quarter Can't Save Chips, Apple Becomes the Low-Spend Shelter

The Closing Snapshot: A Market Split Along Capex Lines

The semiconductor sector is one of the cleanest tells in the July 24 closing snapshot. The VanEck Semiconductor ETF (SMH) fell 3.27% to $561.19^[1], dragging the Technology Select Sector SPDR (XLK) down 1.44%^[1] and the Invesco QQQ Trust (QQQ) down 1.12% to $684.23^[1]. Meanwhile, the SPDR Dow Jones Industrial Average ETF (DIA) rose 0.48% to $518.76^[1] and the SPDR S&P 500 ETF (SPY) edged up 0.10% to $738.93^[1].

The split was not subtle. Financials (XLF) gained 0.86%, energy (XLE) rose 0.40%, and healthcare (XLV) advanced 0.70%^[1] — defensive and cyclical value sectors firmed while the AI-exposed growth sleeve sold off. The Russell 2000 (IWM) slipped 0.31%^[1], neither fully joining the risk-on rotation nor escaping the growth pullback.

This was the second straight week of losses for the S&P 500 and Nasdaq^[2], with the decline driven less by macro headlines and more by a fundamental question the market is now asking about every earnings print: how much cash is AI infrastructure consuming, and when does it start returning?

Intel’s Paradox: Record Revenue, 8% Sell-Off

Intel reported its fastest quarterly revenue growth in nearly 15 years on Thursday after the close. Q2 revenue climbed 25% year over year to $16.1 billion, well above the $14.42 billion consensus^[3]. Adjusted earnings came in at 42 cents per share, nearly double the 21-cent estimate^[3]. The Data Center and AI Group surged 59% to $6.3 billion^[3], and gross margin rebounded to 42% from just 2.5% a year earlier^[3].

The stock fell 7.89% to $92.32^[4].

The disconnect matters. Intel also guided Q3 earnings of 38 cents per share on $15.8–$16.8 billion in revenue, comfortably ahead of the 27-cent, $15.1 billion consensus^[3]. By any conventional measure, this was a beat-and-raise. So why did the stock collapse?

The answer is capital expenditure. Intel said capex will rise meaningfully next year as it expands its foundry business^[3]. The foundry division generated $5.8 billion in quarterly revenue, up 31% year over year^[3], but investors are still waiting for a major flagship foundry customer to validate the strategy against TSMC^[3]. The company has signed 10 long-term agreements for server processors^[3], but that is supply-locking, not yet proof that the foundry can win external business at scale.

The read-through is that in the current tape, revenue growth alone no longer clears the bar. The market is pricing the cost of building the capacity behind that growth — and finding it expensive. Intel’s stock has now fallen roughly 28% this month, though it remains up more than 170% year to date^[3].

The sell-off was not isolated to Intel. Across the semiconductor complex: AMD fell 3.29% to $521.95, Broadcom (AVGO) dropped 2.69% to $381.92, and NVIDIA (NVDA) slipped 0.92% to $206.84^[4]. Meta Platforms (META) declined 1.80% to $595.19^[4]. These are the names whose valuations are anchored to AI infrastructure spending — and that spending is now being scrutinized for its return profile.

The Magnificent Seven’s $797 Billion Reassessment

The semiconductor pressure came against a broader backdrop. On July 23, the Magnificent Seven shed roughly $797 billion in market value in a single session — their worst day since the April 2025 tariff selloff^[5]. Tesla fell 15% and Alphabet dropped 7.1%^[5] as AI capital expenditure projections from Alphabet’s earnings sparked fresh doubts about the sustainability of the AI spending cycle^[5].

The concern is specific and quantifiable. Alphabet’s second-quarter free cash flow was negative $5.86 billion despite 24% revenue growth and 82% cloud growth^[6]. The company’s property, plant, and equipment purchases consumed 35.1% of sales^[6]. When a company growing revenue 24% is burning cash, the market’s question about when the AI infrastructure cycle begins paying for itself is not abstract.

Apple’s Shelter Premium

If Intel and Alphabet represent the cost side of the AI buildout, Apple represents the shelter. AAPL rose 3.53% to $333.02^[4] — the standout performer among megacap technology on a day when most of the AI-exposed complex sold off.

The logic is visible in the cash-flow math. Over the most recent six-month period, Apple generated $82.63 billion in operating cash flow and spent just $4.34 billion on property and equipment, leaving $78.28 billion in free cash after PP&E — 94.7% of operating cash flow^[6]. Alphabet generated nearly identical operating cash flow ($84.86 billion) but spent $80.60 billion on PP&E, leaving just $4.26 billion — 5.0% of operating cash flow^[6]. The 18-fold gap in post-capex cash is the market’s case for Apple as a “port in the storm”^[6].

Baird analyst William Power raised his Apple price target from $310 to $330 and kept an Outperform rating, citing Apple’s low capex intensity as a shelter from the AI spending storm^[6]. The stock traded above that new target during Friday’s session^[6].

The premium has a cost. Apple now trades at 40.3 times trailing earnings, versus Alphabet’s 16.0 times^[6]. Power himself acknowledged the valuation is rich against Apple’s historical range^[6]. The March quarter set a demanding comparison base — revenue rose 17%, with iPhone sales up 22% and services gaining 16%^[6] — and Apple reports fiscal Q3 results on July 30^[7]. A weaker outlook could unwind Friday’s shelter premium quickly.

Two interpretations sit side by side. The first says the market is correctly repricing AI infrastructure risk: the buildout is real, demand is genuine (Intel’s 59% data center growth confirms it), but the capital intensity is extreme and the payback is unproven. The second says this is the kind of capex anxiety that accompanies every major infrastructure cycle — the railroads, telecom, the cloud buildout — and the companies spending now will be the ones earning outsized returns when the cycle matures. The truth is probably somewhere between, and earnings season over the next two weeks will produce the evidence for each side.

The Oil Relief Valve

While the AI capex question split the tech complex, the Dow and S&P found support from a pullback in oil prices. Crude retreated from above $100 per barrel^[8] on reports that China was pushing for an end to the US-Iran conflict^[8], offering the first hint of a diplomatic off-ramp in a week of escalating hostilities.

The backdrop remains volatile. The US launched fresh strikes on Iran overnight^[8], and President Trump threatened a “massive attack”^[8] after Houthi forces struck two Saudi oil tankers in the Red Sea^[8]. Iran rejected a temporary ceasefire until its Strait of Hormuz demands were met^[8]. Both the Strait of Hormuz and the Bab el-Mandeb Strait remain under simultaneous threat^[8].

The oil retreat was enough to let the Dow gain ground — energy stocks (XLE) were modestly positive^[1], and oil majors XOM and CVX were roughly flat^[4] — but it was a cooling of the premium, not a resolution of the risk. The Middle East remains the market’s primary exogenous uncertainty, and any re-escalation would quickly reverse the relief trade.

The Macro Backdrop: Growth Holding, Sentiment Slipping

The macro picture is not flashing recession. Real GDP is growing at 2.66% year over year^[9], unemployment is stable at 4.2%^[9], and high-yield credit spreads remain tight at 2.69%^[9] — the credit market is not pricing stress. The VIX at 16.64^[9] is low by historical standards, though the index has risen from its recent trough.

The tension is in the consumer. Sentiment fell to 44.8, down 14% year over year^[9], a level that historically correlates with spending caution. With CPI still at 3.46%^[9] and the 10Y yield at 4.55%^[9], real disposable income is being squeezed. The macro analogs flagged by the snapshot — mid-2006 and October 2007^[9] — are periods where growth held but the foundation was beginning to crack. Neither analog guarantees a downturn, but both are reminders that soft-landing outcomes are not the base case just because the current numbers look stable.

What to Watch Next

• Apple earnings (July 30): The shelter premium is now the market’s most crowded trade on the capex-shelter thesis. Apple reports after the close on July 30^[7]. The March quarter’s 17% revenue growth^[6] sets a high bar. A weak iPhone or services outlook could reverse the 3.5% gain and test whether the low-capex narrative is enough to hold the 40x multiple.

• Magnificent Seven earnings cascade: With Alphabet and Tesla having already triggered the $797 billion repricing^[5], the remaining megacap earnings — Microsoft, Amazon, Meta, NVIDIA — will each be weighed on the same capex-versus-return axis. Microsoft was flat on Friday^[4], suggesting the market is reserving judgment.

• Middle East escalation chain: Oil’s retreat from $100 enabled Friday’s Dow rally^[8], but Iran’s rejection of the ceasefire^[8] and Houthi attacks on Red Sea shipping^[8] keep the chokepoint risk live. Any disruption to Strait of Hormuz traffic would put oil back above $100 and shift the entire market’s risk pricing.

• Semiconductor demand signals vs. capex: Intel’s 59% data center growth^[3] and 10 long-term supply agreements^[3] confirm AI demand is real. The question the market is pricing is whether the capital required to meet that demand earns an adequate return. Citi called the chip pullback a buying opportunity^[2] — a view that will be tested if capex guidance continues to rise across the complex.

• Consumer sentiment trajectory: At 44.8^[9], sentiment is a leading indicator that the hard data has not yet confirmed. If July spending data softens, the growth-holding narrative weakens and the rotation from growth-to-value could accelerate.