After the US stock market crash, has the century-long peak arrived? Is the AI bubble about to burst?

Today, SK Hynix's earnings fell short of expectations, triggering another猛烈 sell-off in Asian tech stocks and a bloodbath in the market.

Looking back at this adjustment:

• The Nasdaq has fallen about 8%-10% from its highs;
• The Philadelphia Semiconductor Index has dropped about 25% from its June peak;
• South Korea's KOSPI index plunged 10.84% in a single day, a 35.81% retreat from its historic high.

But just as chip stocks were in a bloodbath, Apple reclaimed the world's largest market cap, and the Dow Jones hit a new record high. It appears funds haven't completely left the US stock market; they are simply rotating out of the crowded AI hardware trade into assets with more stable cash flows and less valuation pressure.

This has split the market into two camps: one shouting "The AI bubble has finally burst," while the other insists "This is just a healthy shakeout; the AI narrative hasn't changed."

So, the real question is:

Is the US stock market facing a "once-in-a-century top," or is it still climbing the "wall of worry" within a bull market? Is the AI bubble about to fully burst, or is it merely undergoing a temporary valuation repricing?

⬇️ Below, watch the bulls and bears debate.

Bullish View: The AI Narrative is Intact, This is Not a Bubble Burst

Argument 1: Cathie Wood @CathieDWood | Founder of ARK Invest | XHunt Rank: 170

Cathie Wood believes the market is still climbing a typical "wall of worry."

Real market tops usually don't occur when everyone is full of doubt, but are born from extreme optimism—when investors commonly believe "the sky's the limit," that's when risk can truly spiral out of control.

The current market is full of controversy, panic, and skepticism, which doesn't resemble a classic final top.

Core Judgement: This is likely a normal correction within a bull market. The long-term logic of disruptive innovation hasn't changed; real tops usually appear when consensus is extremely aligned.

Argument 2: RamenPanda @IamRamenPanda | XHunt Rank: 3994

RamenPanda shifts focus from AI hardware to the software and application layer.

The market worries about overcapacity in data centers and computing infrastructure, but these infrastructures ultimately need to generate returns through software, platforms, and commercial applications.

The internet era also experienced a phase of infrastructure overbuilding. Cisco was the early star, but the true long-term winners were Google, Microsoft, and Amazon that rose later.

The AI era might follow a similar path: hardware investment overheats first, then value gradually shifts to software and applications.

Core Judgement: Infrastructure overbuilding might be a necessary stage of technological revolution. The current situation seems more like a transition from hardware build-out to application monetization, not the end of the AI era.

Argument 3: 投資 TALK 君 @TJ_Research | XHunt Rank: 9918

After SK Hynix's earnings report, its stock price initially opened about 8% lower but quickly rebounded.

投資 TALK 君 believes this contrasts with the "sell the news" reaction after Micron's earnings, potentially signaling a short-term bottom.

If bad news is already fully priced in, even disappointing earnings might see selling pressure exhaustion. More importantly, there are currently no signs of systemic weakness in AI demand guidance.

Core Judgement: Short-term sentiment has been vented, but AI demand hasn't been disproven yet. The market needs data confirmation, not more panic.

Argument 4: Nico 投資有道 @tychozzz | XHunt Rank: 13869

Nico believes the biggest lesson from this crash is: no matter how good the fundamentals or low the P/E ratio, when assets are at high levels and the market decides to compress valuations, stock prices can fall sharply. The memory and storage sector is the prime example.

However, "valuation compression for stocks" doesn't equal "disproving the AI industry logic."

The Nasdaq's forward P/E ratio is now near levels seen at the end of March this year. Instead of panic selling, it's more important to control leverage, set target prices for profit-taking, and hold indices and quality companies that can deliver real earnings.

Core Judgement: Sentiment can quickly compress valuations, but ultimately, earnings determine stock prices. The AI narrative hasn't changed; it might just be a matter of time before quality assets return to previous highs.

Argument 5: 美股仙人 @hanking66 | XHunt Rank: 44611

美股仙人's view is more direct: Don't sell stocks at the moment of maximum panic.

In his view, this decline shows the emotional characteristics of a "final washout"—despair, suffocation, breached defenses, with investors interpreting any news most pessimistically.

The fact that many stocks were heavily sold off before their earnings reports suggests that pessimistic expectations may have already been priced in.

Core Judgement: Buy the dip. Endure the most extreme emotional phase; opportunities may belong to those who still have the capacity to buy.

Bearish View: Cracks in AI's Creditworthiness Have Appeared, the Bubble is Deflating

Argument 1: ZeroHedge @zerohedge | Prominent Macro Analysis Media | XHunt Rank: 200

ZeroHedge believes the market is re-entering a pricing phase where "circular financing is actually quite bad."

Multiple danger signals have already appeared:

• The Nasdaq 100 is approaching a 10% technical correction;
• Semiconductor and momentum factor trades have been hit hard;
• Credit cracks are beginning to show among AI-related companies;
• Deals involving Nvidia and OpenAI are raising concerns about circular financing again;
• China's low-cost models and open-source tools continue to pressure profit expectations.

If major AI companies invest in, purchase from, and generate revenue for each other, it becomes difficult for the market to discern how much comes from genuine end-user demand versus how much is just capital circulating within the industry chain.

Core Judgement: The credit and financing expansion in the AI industry is hitting a wall. The bubble's deflation phase may begin sooner than most investors expect.

Argument 2: Phyrex @PhyrexNi | XHunt Rank: 784

Phyrex focuses more on the leverage risk behind the US stock market.

Currently, US investors have borrowed approximately $1.53 trillion to buy stocks, while broker-dealer net credit balances have fallen to about negative $1.061 trillion. Margin debt increased by about $86 billion in a single month and has been rising for three consecutive months.

This means the US stock market isn't just overvalued; its dependence on borrowed funds is also increasing. Once incremental funding slows, the mechanical buying that drove the rally could turn into mechanical selling.

This risk also transmits to Crypto. During cross-market deleveraging, BTC is more likely to be sold first as a high-beta risk asset, rather than immediately rising independently as "digital gold."

However, short positions in US stocks are also high—nearly 6% for the Russell 3000 and around 3.5% for the S&P 500—leaving room for a potential short squeeze.

Core Judgement: The real risk might not be the AI technology being disproven, but the simultaneous occurrence of high valuations, high leverage, and slowing capital inflows. Once deleveraging starts, both US stocks and BTC could face mechanical selling.

Argument 3: Peter Schiff @PeterSchiff | Economist, Founder of SchiffGold | XHunt Rank: 1486

Peter Schiff argues that AI itself is a real technology, but AI-related stocks have formed a serious bubble.

China's low-cost models, open-source technology, and increasing industry competition will continue to pressure the profit margins and moat expectations of US AI companies.

If massive capital expenditures cannot generate matching cash flows, more valuation "air" will still need to be squeezed out.

Core Judgement: AI technology is real, but the AI stock bubble is equally real. The bubble may have already burst, and what lies ahead is a sustained period of deflation.

Argument 4: CryptoPainter @CryptoPainter | XHunt Rank: 2207

CryptoPainter views the crash in the South Korean stock market as a stress test for the AI infrastructure narrative.

He doesn't deny the demand for AI computing power and data centers, but he is wary of the circular investment, excessive financing, and high leverage built around this narrative.

Therefore, this adjustment may not replicate the complete industry collapse of the 2000 era, but it could mirror the violent deleveraging seen in the Korean market: the technological direction remains valid, but related asset prices correct sharply due to fragile valuation and financing structures.

For Crypto, if BTC breaks down again, one of the most likely external triggers would be a continued decline in US stocks.

Core Judgement: AI technology being real doesn't mean related asset prices are reasonable. Compared to the industry going to zero, a more realistic risk is the deleveraging of financing and leverage, transmitted to BTC via US stocks.

Argument 5: 海倫子 Hellen @peng_hellen | XHunt Rank: 40426

Hellen compares today's AI frenzy to the British railway bubble and the internet bubble.

Railways and the internet truly changed the world, but capital frenzy still led to excessive infrastructure construction. Some projects incurred heavier losses the more they invested and operated.

This is similar to the dilemma faced by some large model companies: more users mean higher computing and inference costs, but revenue may not cover the investment.

The internet didn't disappear after the bubble burst, but the Nasdaq still fell for 31 consecutive months, correcting about 78% from its peak.

Core Judgement: Great technology and massive bubbles can coexist. AI might change the world, but related asset prices can still severely deviate from true value and demand.

Summary of Both Sides

The Bulls Argue

The current crash is mainly due to the unwinding of crowded trades and sentiment-driven valuation compression. No data yet proves that AI demand has systemically collapsed.

New competitors and low-cost models won't necessarily destroy US tech giants; they might lower the entry barrier for AI usage and expand the application market.

The AI narrative is not over. The market can still advance while climbing a "wall of worry."

The Bears Argue

The moats of AI companies are shallower than the market assumes.

High valuations, high capital expenditure, leverage, and circular financing have propped up the past rally. Once revenue and cash flows can't match the investment, the entire valuation system faces repricing.

Technological revolutions don't disappear, but investment bubbles can burst prematurely.

Finally

What US tech stocks are experiencing is not a wholesale rejection of the long-term AI thesis, but the market beginning to recalculate a fundamental question:

If return on capital expenditure is insufficient, circular financing recedes, and China's low-cost models continue to pressure prices, how much moat is left for US tech leaders?

The Nasdaq correction, storage stocks being halved, and the plunge in Korean chip stocks are all essentially a concentrated release of this specific worry.

But declaring that the "once-in-a-century top has arrived" is perhaps premature.

The true watershed is not the current sentiment-driven valuation compression, but what happens over the next few quarters and beyond: can AI revenue and free cash flow match the massive capital expenditures? Can industry chain demand expand from circular procurement among giants to truly paying end-users?