June 23, 2026, marked what many are calling "Black Tuesday" in global capital markets. South Korea’s KOSPI Index closed with a dramatic plunge of 910.71 points, down 9.99% to 8,203.84. The sell-off triggered a circuit breaker after an 8% drop, halting trading for 20 minutes. Samsung Electronics and SK Hynix both tumbled over 12%, becoming the epicenter of the rout. Japan’s Nikkei 225 ended the day down 3.55% at 69,788.38. Nasdaq 100 futures fell 2.5% in pre-market trading. Against this backdrop of broad risk-asset selloffs, Bitcoin dropped toward $62,000.
This turmoil was not an isolated incident, but rather the culmination of long-building structural pressures finally erupting in the secondary markets.
Why South Korea’s Stock Market Became the Epicenter of the AI Compute Bubble Burst
The KOSPI’s 9.99% single-day drop was a historic event—ranking as the fifth-largest daily percentage decline in the index’s history.
South Korea’s outsized losses are closely tied to the unique structure of its stock index. Samsung Electronics and SK Hynix, two giants in the memory chip sector, together make up about 50% of the KOSPI’s weighting. These companies are at the heart of the global AI compute infrastructure supply chain, and their stock performance directly reflects global expectations for AI capital expenditures. When investors began to reassess the return on AI investments, these two stocks naturally became the primary targets for capital outflows.
On the day, Samsung Electronics plunged 12.31%, while SK Hynix dropped 12.47%. South Korea’s financial regulators sharply criticized leveraged ETFs tracking these two companies, stating that such products "serve little purpose other than allowing securities firms to profit at the expense of retail investors." The amplifying effect of leveraged products further steepened the market’s decline.
Why $725 Billion in CapEx by the Big Four Cloud Providers Is Raising Serious Doubts
The root driver of this sell-off is mounting skepticism over the return on the massive capital expenditures by the four major cloud providers.
According to Goldman Sachs’ June 2026 forecast, the four hyperscale data center operators—Alphabet (Google), Amazon, Microsoft, and Meta—are projected to spend a combined $725 billion in capital expenditures in 2026, up 77% from $410 billion in 2025. For context, this figure was only about $250 billion in 2024. In just three years, the Big Four’s CapEx has nearly tripled.
However, CapEx growth has clearly outpaced revenue growth. Bank of America estimates that in 2026, hyperscale cloud providers’ CapEx will consume about 90% of their operating cash flow, up from 65% in 2025. Barclays expects Alphabet and Meta’s free cash flow to plunge nearly 90% this year.
Even more concerning is the fragility of their financing structure. In 2025, these hyperscale cloud companies issued over $100 billion in bonds—four times the $28 billion annual average from 2020 to 2024. As of October 2025, AI-related debt had ballooned to $1.2 trillion, accounting for 14% of JPMorgan’s U.S. Liquidity Index, surpassing that of major U.S. banks.
When a company needs to borrow an amount equivalent to 90% of its operating cash flow just to maintain CapEx, the market’s focus inevitably shifts from "growth narratives" to "return validation."
How Falling Compute Rental Prices Are Undermining the "Compute Scarcity" Narrative
For the past three years, the AI industry’s core narrative has followed a simple logic: the scarcer the compute, the more justified the CapEx; the higher the CapEx, the higher the valuation; the higher the valuation, the easier it is to raise funds. This self-reinforcing cycle went largely unchallenged.
But by mid-2026, every link in this chain is under stress. The most direct signal comes from the compute rental market: the hourly rate for Nvidia’s flagship AI chip, the B200, dropped from $6.11 on May 30 to $4.22 by June 21.
The sustained decline in compute rental prices signals that the "compute scarcity" narrative is being eroded by real-world shifts in supply and demand. If scarcity were still a reality, prices would remain firm, providing justification for ongoing CapEx. As prices begin a downward trend, the fundamental logic supporting lofty valuations starts to crack.
At the same time, a rare divergence has emerged between spot and forward prices for compute rentals, revealing a deep conflict between short-term oversupply and long-term demand expectations. Tech giants are tightening AI budgets, power and engineering delivery capabilities are hitting physical limits, and capital markets are now scrutinizing every AI company through the lens of ROI.
Why Bitcoin Is Falling in Tandem with Nasdaq Futures
On June 23, Bitcoin briefly surged to $65,500 in early trading before plunging to $62,900, down about 2% over 24 hours. Gate’s market data shows Bitcoin traded in the $63,900–$64,300 range that day, with an intraday swing between $62,000 and $65,500.
Bitcoin’s drop alongside Nasdaq futures highlights its risk-asset characteristics in the current macro environment. Although the 40-day correlation between Bitcoin and the Nasdaq had dropped to zero by early June 2026, on a broader scale, Bitcoin’s correlation with the Nasdaq 100 Index remains around 0.45—higher than its 10-year average. This means Bitcoin still struggles to decouple from tech stocks during systemic risk events.
The transmission path of this sell-off is clear: U.S. tech giants weakened sharply on Monday; Asian markets, including Japan and Korea, followed with steep declines; pre-market, Nasdaq 100 futures fell 2.5%; and Bitcoin then dropped to around $62,000. This forms a classic chain: "AI compute stocks → global risk assets → Bitcoin."
It’s also worth noting the shift in market drivers. For weeks, Bitcoin’s price swings tracked every twist in Middle East geopolitics. Now, as the U.S. and Iran move toward a peace roadmap, the driving force has shifted to the same AI-fueled tech trade that pushed stocks to record highs. As that trade falters, crypto follows suit.
How Crypto Mining Firms Face Both Peril and Opportunity Amid AI Compute Oversupply
The reversal in AI compute supply and demand is fundamentally reshaping the survival logic for crypto mining firms.
Since May, Bitcoin network hash rate has dropped by 145 EH/s—the first such contraction in six years—as miners redirect electricity and sell BTC to finance AI data center construction. In Q2 2026, hash rate fell 5.8% quarter-over-quarter to 1,004 EH/s. Electricity now accounts for 70%–90% of mining operating costs, and competition from AI data centers is making cheap power harder to secure.
From a revenue perspective, AI data centers can generate $200–$500 per megawatt, while Bitcoin mining yields only $57–$129 per megawatt. This income gap makes the financial logic of reallocating power to AI workloads obvious for miners.
However, this transition comes with significant funding constraints. Crypto mining firms face a near-term funding gap of about $5 billion as they attempt to convert power assets into AI data centers. Meanwhile, falling compute rental prices are squeezing the profit window for this transition. If prices continue to drop, miners’ AI transformation will face the challenge of "front-loaded investment, delayed and uncertain returns."
Why Micron’s Earnings Report Is a Key Test for Risk Assets
All eyes are now on U.S. memory chip giant Micron Technology, which is set to release its quarterly earnings on June 24 (Wednesday). This report is widely seen as a crucial test of whether AI spending can continue to support current market valuations.
Micron’s earnings are so important because they directly reflect demand for memory chips—the most critical hardware component in AI compute infrastructure. The sharp declines in SK Hynix and Samsung Electronics are, in essence, a pre-pricing of Micron’s results. If Micron disappoints, it will further fuel doubts about the returns on AI capital expenditures; if the results beat expectations, the oversold semiconductor sector could see a short-term reprieve.
Bloomberg strategists noted, "Recent risks for regional chip stocks include increasingly unstable market structure and Micron’s earnings report after the U.S. close on Wednesday. Concerns are mounting over whether U.S. hyperscale cloud providers’ unprecedented AI infrastructure investments are prudent."
From "Open Bar" to "Rationing": A Structural Turning Point for the AI Industry
"Black Tuesday" on June 23, 2026, may well be remembered as the dividing line between the first and second halves of the AI industry cycle.
The first half was defined by an "open bar" mentality—unlimited CapEx, soaring valuations, and easy financing. The second half is marked by "rationing"—capital demanding returns, valuations facing scrutiny, and financing becoming constrained.
Falling compute rental prices, tech giants collectively tightening AI budgets, and the exposure of physical limits in power and engineering delivery—all three cracks are opening at once, pushing the AI industry into a new phase. Capital markets are now judging every AI company by ROI, no longer relying solely on the "compute scarcity" narrative to justify valuations.
For the crypto market, this means Bitcoin’s status as a "risk asset" will be tested throughout the unwinding of the AI compute bubble. Whether Bitcoin can maintain its "digital gold" narrative in a risk-off environment will largely depend on macro liquidity conditions and crypto market fundamentals—including ETF flows, stablecoin regulatory developments, and post-halving supply dynamics.
Conclusion
The global market turmoil of June 23, 2026, essentially represents a concentrated reckoning with the returns on more than a trillion dollars in AI capital expenditures over the past four years. The KOSPI circuit breaker, the Nasdaq futures plunge, and Bitcoin’s pullback all reflect the same logic playing out across different asset classes. As compute rental prices fall, cloud providers’ free cash flow comes under pressure, and AI-related debt balloons, the "compute scarcity" narrative can no longer support current valuation frameworks.
For crypto market participants, understanding the evolution of the AI compute bubble is fundamentally about understanding Bitcoin’s risk-asset pricing logic. Until the returns on AI CapEx are validated, Bitcoin’s high correlation with tech stocks is likely to persist.
Frequently Asked Questions (FAQ)
Q: Why does the AI compute bubble burst affect the price of Bitcoin?
Bitcoin is still priced as a risk asset in the current market environment. When investors question the returns on AI capital expenditures, capital flows out of high-beta tech stocks and risk assets—including Bitcoin, which naturally comes under similar selling pressure. This latest sequence—from "U.S. tech stocks weakening → Asia-Pacific markets plunging → Nasdaq futures tumbling → Bitcoin pulling back"—clearly illustrates this linkage.
Q: Which companies are included in the $725 billion CapEx by the Big Four cloud providers?
This refers to Alphabet (Google), Amazon, Microsoft, and Meta—the four U.S. hyperscale cloud providers. According to Goldman Sachs’ June 2026 forecast, these companies’ combined CapEx will reach $725 billion in 2026, up 77% from $410 billion in 2025.
Q: What does a drop in compute rental prices mean?
Compute rental prices are the most direct indicator of AI compute supply and demand. The hourly rental rate for Nvidia’s B200 chip fell from $6.11 at the end of May to $4.22 in late June, signaling that compute supply is outpacing demand growth. This directly undermines the "compute scarcity" narrative—which has been the core logic supporting high valuations of AI-related assets over the past three years.
Q: How are crypto mining firms affected by AI compute oversupply?
On one hand, competition for electricity from AI data centers is driving up miners’ power costs—power now accounts for 70%–90% of mining operating expenses. On the other hand, miners face a roughly $5 billion funding gap as they attempt to convert power assets into AI data centers. If compute rental prices continue to fall, profit expectations for miners’ AI pivot will be further compressed.
Q: Will Bitcoin continue to fall in tandem with tech stocks?
Bitcoin’s correlation with the Nasdaq 100 currently stands at about 0.45, above its 10-year average. Until the market validates the returns on AI capital expenditures, Bitcoin’s high correlation with tech stocks is likely to persist. Key variables to watch include Micron’s earnings, the Federal Reserve’s monetary policy trajectory, and capital flows within the crypto market itself.
