US tech stocks suffered a sharp plunge, and the Philadelphia Semiconductor Index fell into a bear market: is AI faith collapsing?

On the early morning of July 28, 2026 Beijing time, the three major U.S. stock indexes closed mixed. The Dow Jones Industrial Average closed at 52,210.08, up 0.51%; the S&P 500 inched up 0.02% to 7,413.18; the Nasdaq Composite closed at 24,932.08, down 0.18%, recording its fourth consecutive session of declines.

Sector-level divergence masked severe internal rifts. Traditional tech giants represented by Apple and Microsoft rose against the trend, while the semiconductor sector saw broad-based selling. The Philadelphia Semiconductor Index (SOX) closed down 2.23% to 11,554.88. This closing level marked a ten-week low since May 20. Compared with the historical all-time high of 14,655.29 hit on June 22, SOX plunged 3,100.41 points in just five weeks, bringing its cumulative pullback to 21%, officially sliding into a technical bear market.

NVIDIA (NVIDIA) the AI chip leader fell 4.99% on the day to $196.51, its largest single-day decline since June 5. Its total market value shrank by nearly $250 billion in a single day, falling to $4.76 trillion.

This chip-stock-driven selloff wave is pushing one question to the forefront: Is the market’s faith in AI undergoing a fundamental shake-up?

Why the Nasdaq fell four straight days while the Dow rose

The Nasdaq index has declined for four consecutive trading days, but the Dow Jones Industrial Average logged a two-day gain over the same period. This kind of divergence itself carries an important market signal.

The Dow’s rise was mainly driven by traditional industrial and energy sectors. Although the energy sector weakened that day due to the drop in oil prices, sectors that benefit from lower fuel costs—such as airlines and tourism—moved up collectively. Among the 11 sectors in the S&P 500, both the telecom and consumer staples sectors posted gains of more than 1%.

The Nasdaq’s decline, meanwhile, was entirely dragged down by technology stocks—especially the semiconductor sector. The VanEck U.S. Software? (WIND) America Technology “Seven Giants” index fell 0.33% that day. Excluding AI-related stocks, the S&P 500 actually recorded positive returns of 0.80%, clearly showing: the AI sector has become the only core factor dragging down the broader market.

This “Dow up, Nasdaq down” pattern, in essence, reflects capital rotating away from high-valuation tech stocks and into defensive sectors. The market is stress-testing the one-way rally logic driven by AI over the past two years.

What it means that the Philadelphia Semiconductor Index has fallen into a bear market

The definition of a technical bear market typically refers to an index pulling back by 20% or more from a recent high. SOX’s drop from the historical peak of 14,655.29 on June 22 to 11,554.88 already represents a 21% decline, clearly meeting this standard.

But more than the “bear market” label, what deserves greater attention is the speed and breadth of the decline. SOX surged more than 80% cumulatively in the first half of 2026, and the second quarter alone rose 88%—marking the strongest single-quarter performance in the index’s history. After such a rapid rally, a 21% pullback within five weeks indicates that this kind of sharp two-way volatility suggests the semiconductor sector’s pricing has moved away from fundamental anchors, and is increasingly driven by market sentiment and positioning battles.

From a technical perspective, SOX is currently standing above the key support level around 11,200. Based on LSEG Workspace data, if SOX sees a decisive closing break below 11,200, there will be no effective support until the region around the 200-day moving average. This moving average currently closely overlaps with the long-term uptrend line, forming a more decisive technical line of defense.

The deep logic behind NVIDIA’s near 5% one-day drop

NVIDIA’s near-5% decline on the day was not an isolated event, but the result of multiple pressures stacking together.

The immediate trigger was that market doubts about NVIDIA’s “circular financing” model resurfaced and intensified. In recent weeks, NVIDIA disclosed multiple large-scale partnership plans, including supply-chain cooperation with SK Group exceeding $500 billion, as well as plans to provide up to $250 billion in compute rental guarantee for OpenAI.

Bloomberg Macro strategist Michael Ball noted that these massive infrastructure agreements reignited market concerns about the “revolving cash flow” logic behind AI capital expenditure. The core issue is that NVIDIA’s revenue growth is highly dependent on the financing capacity of downstream customers—funds circulate within the industrial chain: NVIDIA sells chips to cloud providers, cloud providers finance and build data centers, and part of the financed funds then flows back to NVIDIA to buy more chips. If the financing environment changes, the entire chain of AI spending faces contraction risk.

NVIDIA’s credit derivatives market also sent warning signals. Its credit default swap (CDS) spread jumped by 14 basis points in a single day, marking the largest single-day increase in history. This means the market’s pricing of NVIDIA’s credit risk is rising rapidly.

Goldman Sachs analyst Chris Hussey pointed out that the S&P 500’s lagging performance over the past two months stems primarily from market doubts about the sustainability of AI infrastructure investment, not macro factors such as oil prices or interest rates.

Why the plunge in storage chips carries even stronger signals

If NVIDIA’s decline reflected the market’s concerns about the AI chip leader, then the broad breakdown in the storage chip sector reveals the vulnerability of the entire AI hardware supply chain.

Sandisk plunged 11.02% that day. SK Hynix U.S. depositary receipts fell 7.47% to $143.02; it broke below the IPO issue price of $149 after only 12 trading days on the market. Micron Technology was down more than 7% at one point during the day, narrowing to a 2.25% close into the end. Western Digital and Seagate Technology both fell more than 4%.

Storage chips are the underlying hardware of AI infrastructure, and their prices and demand directly reflect the true pace of data center construction. A breakdown lower across the storage segment may mean the market is starting to doubt whether the data center buildout pace for AI has been over-anticipated.

It is also worth noting that storage chips were one of the strongest-performing sectors in the first half of 2026. The rapid reversal from leading to lagging is often a standard path for a sector bubble to burst.

Why the AI “circular financing” model triggers a trust crisis

To understand the core of this round of selloff, we need to go back to a fundamental question: what actually supports the capital expenditures of AI infrastructure?

The narrative from the past two years has been: cloud providers (Microsoft, Google, Amazon, etc.) pour in massive capital expenditures to build AI data centers. These expenditures translate into revenue for chipmakers such as NVIDIA. The chipmakers’ performance growth then supports their stock prices’ continued rise. This is a self-reinforcing positive loop.

But the fragility of this loop lies in: cloud providers’ capital expenditures are not accumulated naturally from free cash flow, but rely heavily on debt financing. Microsoft’s 2026 capital expenditure is expected to exceed $200 billion, and Amazon is roughly about $200 billion as well. Alphabet raised its 2026 capital expenditure forecast to a maximum of $205 billion and reported its first-ever negative free cash flow. Meta increased its full-year capital expenditure guidance to between $125 billion and $145 billion.

When the growth rate of capital expenditures continues to exceed the growth rate of free cash flow, debt leverage keeps building up. Once financing costs rise or market risk appetite declines, the entire chain could collapse in reverse—which is exactly the scenario the market is currently pricing in.

Thomas Martin, senior portfolio manager at the U.S. asset manager Globalt Investments, said directly that the AI bubble is showing signs of “deflating,” and traders are reshuffling their positions.

Does Apple reclaiming the top spot by market value signal a style shift

While the market collectively sold off AI hardware stocks, Apple’s share price rose 1.17% on the day to $336.91, bringing its total market value to $4.95 trillion, again surpassing NVIDIA to become the world’s most valuable listed company.

This change of the “stock king” carries important symbolic meaning. NVIDIA has held the top spot since June 2025 when it took it from Microsoft and briefly touched $5 trillion in market value in October 2025. In 2026 so far, NVIDIA’s share price is up only about 4%—5.5%, while Apple is up about 24%.

The divergence between Apple and NVIDIA is, in essence, a market pricing difference between two AI strategies. Apple is not blindly spending heavily on artificial intelligence; it is more inclined toward “renting compute power” rather than building AI compute infrastructure itself. This relatively restrained capital expenditure strategy is winning favor under current market conditions.

Meanwhile, Google rose 2.13% and Microsoft rose 1.94%, and the software sector saw broad gains. Money is rotating away from the semiconductor upstream, which has high valuations and relies on long-dated expectations, and toward consumer tech and enterprise software with stronger earnings certainty and more stable cash flows.

The concept of “the seven tech giants” as a whole is losing its explanatory power. Apple and Microsoft have a spread of more than 40 percentage points in performance this year. Tesla is down more than 30% year-to-date, making it the worst-performing single stock among the seven. This internal split indicates that the market is moving from “tag-based investing” to “fine-grained pricing.”

Key variables to watch next

Does this adjustment in tech stocks mean the end of the long-term AI narrative? It’s still too early to draw conclusions. But several key variables are worth continuous tracking.

First, cloud providers’ earnings reports and capital expenditure guidance. Microsoft, Amazon, Meta, and Apple will report earnings in sequence. The relationship between AI capital expenditures and free cash flow will become the core basis for the market to judge “circular financing” risk.

Second, SOX’s technical support. Whether the 11,200 level can hold will determine whether the semiconductor sector will trigger broader systemic selling.

Third, changes in the financing environment. The trajectory of U.S. Treasury yields and credit spreads will directly affect the cost and feasibility of debt financing for AI infrastructure.

Fourth, the pace of AI commercialization. If AI-related revenue cannot be realized quickly enough, the payback cycle for capital expenditures will lengthen, increasing valuation pressure further.

Summary

Nasdaq’s four-day slide and the Philadelphia Semiconductor Index entering a bear market mark the market’s first systematic reflection on the AI investment logic. NVIDIA’s nearly 5% one-day drop and the intensification of doubts about “circular financing” reveal the intrinsic fragility of the AI hardware industry chain under debt-leverage-driven dynamics. Apple reclaiming the global market value number-one spot suggests capital is shifting from “telling stories” to “doing the math.”

This is not the end of the AI industry, but it very likely is an important turning point in the AI investment paradigm—from “expand at any cost” to “scrutinize returns.” For market participants, understanding the depth and duration of this shift matters far more than chasing short-term volatility.

Frequently Asked Questions (FAQ)

Q: Why has the Nasdaq index fallen for four straight days?

The main drag behind the Nasdaq’s four-day decline has been sustained selling in the semiconductor sector. The market has questioned the sustainability of AI infrastructure investment, prompting capital to exit high-valuation tech stocks. At the same time, the Dow rose due to support from traditional industrial sectors, showing a clear pattern of sector divergence.

Q: What does the Philadelphia Semiconductor Index falling into a bear market mean?

The Philadelphia Semiconductor Index has pulled back 21% from its historical peak on June 22, officially entering a technical bear market. This means the semiconductor sector has experienced a sharp valuation contraction over the past five weeks, and the market’s expectations for AI hardware demand are being repriced.

Q: Why did NVIDIA’s share price drop nearly 5% in a single day?

NVIDIA fell 4.99% that day, mainly due to market concerns about its “circular financing” model—its revenue growth depends heavily on the ability of downstream customers to finance with debt; if the financing environment changes, the entire chain of AI spending could face contraction risk.

Q: What is the AI “circular financing” model?

“Circular financing” refers to a closed-loop of funds within the AI supply chain: cloud providers finance and build data centers → buy chips such as NVIDIA’s → chipmakers’ performance grows → supports stock prices → further financing expands capacity. The fragility of this model lies in its heavy reliance on external debt financing and the lack of natural support from free cash flow.

Q: Why was Apple able to surpass NVIDIA again to become the world’s top company by market cap?

Apple is up about 24% year-to-date, while NVIDIA is only up about 4%—5.5%. The market has recognized Apple’s comparatively restrained capital expenditure strategy in AI—“renting compute power” rather than building large-scale compute infrastructure itself—which is seen as more sustainable amid a tightening financing environment.

Q: Has the long-term AI narrative already ended?

It’s not yet possible to assert that the long-term AI narrative has ended. However, the market is shifting from a “growth at any cost” phase to a “fine-tuned, cost-effective return” phase. The efficiency of AI capital expenditures, the return cycle, and the quality of cash flow will become the core variables for future pricing.