Is Stock Market Prosperity Back? 'AI Prophet' Faces Leverage Crisis as Semiconductor Shifts to 'Fundamental-Driven Pricing'

The artificial intelligence (AI) investment boom is undergoing a dramatic reshuffle, from AI investor Leopold Aschenbrenner's fund suffering heavy losses due to leverage unwinding, to semiconductor stocks rebounding strongly on fundamental tailwinds. The market is re-evaluating the sustainability of the AI rally. Analysis suggests that the recent downturn was primarily driven by capital reallocation, not weakening demand. As cloud giants continue to invest heavily in AI infrastructure, HBM, memory, and data center supply chains still possess long-term growth momentum.

Situational Awareness, the hedge fund founded by Aschenbrenner—a prodigious investor in his twenties who previously worked on OpenAI's 'Super Alignment' team—suffered a staggering 67% loss in a single month in July.

To protect itself, the fund was forced to offload most of its public market holdings to Citadel, led by Wall Street titan Ken Griffin, and fully unwind its leveraged positions.

Aschenbrenner admitted in a letter to investors: 'This month, we let you down.'

Aschenbrenner's rise was the perfect embodiment of the Silicon Valley narrative. In 2024, he published a 165-page white paper titled 'Situational Awareness,' boldly predicting that Artificial General Intelligence (AGI) would force the world to expand data centers at exponential speed. Entire supply chains—from power equipment and liquid cooling systems to memory chips, optical communication networks, and gas turbines—would be swept into this arms race.

Subsequently, his namesake fund, Situational Awareness, attracted heavyweight capital from figures such as Jane Street and Stripe founders.

Within just two years, the fund's assets under management surpassed $20 billion. By the end of June this year, its year-to-date net return reached 439%, with cumulative returns exceeding tenfold since inception. He was briefly hailed by retail investors as the 'AI investment prophet.'

Even after this sharp correction, Situational Awareness has still delivered an overall return of approximately 80% this year. Analysts point out that what collapsed first was not his investment thesis, but the underlying financing structure.

The Double-Edged Sword of Leverage: Accelerator in Upturns, Noose in Downturns

Analysts note that the core issue behind this event is not new: leverage turns time into an adversary for investors.

When stock prices fall, the value of collateral assets shrinks in tandem, prompting lenders to demand additional margin or force position reductions.

This compels funds to sell their most liquid assets, regardless of whether the original investment thesis remains valid. Aschenbrenner likened this chain reaction to a 'bank run,' where every sign of fragility breeds further fragility.

Notably, the fund ultimately retained its less liquid private holdings, including stakes in Anthropic. What was sold off were not the assets he viewed most pessimistically, but the most easily liquidated public market positions.

Market observers have drawn parallels between this event and the near-collapse of Long-Term Capital Management (LTCM) in 1998—both occurring around the fourth year after their respective tech booms. However, the scale and systemic risk are incomparable. Situational Awareness never posed a threat capable of shaking the financial system. Yet, the underlying logic—'smart money facing a liquidity test'—is identical in both cases.

Calvin Yeoh, co-manager of Merlion Fund under Singapore's Blue Edge Advisors, commented that this wave of selling pressure indeed extended the tense market sentiment of recent weeks. With highly leveraged retail positions in the South Korean market nearly fully unwound, there is unlikely to be a larger-scale selling force ahead.

An Unexpected Turn: After Deleveraging, Semiconductor Stocks Stage an Epic Rebound

Just as the market assumed the AI frenzy was cooling down, the narrative reversed. On Thursday (30th) in U.S. markets and Friday (31st) in Asia, the semiconductor sector delivered a textbook-level comeback.

What fueled this rebound was not merely short covering, but solid fundamental data. South Korea's total exports in July surged 62.8% year-on-year to $98.89 billion, far exceeding market expectations. Semiconductor exports skyrocketed 179%, while computer product exports surged 404%. The primary drivers were AI data center investments, rising memory chip prices, and strong SSD demand.

Exports to both the U.S. and China—the two major markets—also strengthened simultaneously, indicating that global AI hardware demand has not cooled despite the stock price correction.

Cloud giants' earnings reports provided strong corroboration. Microsoft's (MSFT-US) Azure cloud business revenue grew 43%, surpassing the market's expected near-40%. Amazon's (AMZN-US) AWS revenue rose 37% to $42.2 billion, with its backlog jumping from $364 billion to $496 billion. Management even stated that most of its computing capacity through 2027 is already pre-booked by customers. Even with capital expenditure projected to rise to $220 billion next year, supply may still fall short of demand.

Lam Research (LRCX-US), a major semiconductor equipment manufacturer, also set its next-quarter revenue guidance midpoint at $8.1 billion, significantly above the market's estimate of $7.09 billion.

Morgan Stanley, in its report 'Buying the AI Infrastructure Dip,' pointed out that the recent decline in AI infrastructure stocks was primarily due to short-term position adjustments, crowded trades, and technical selling pressure—not deterioration in fundamental demand.

The bank believes that more efficient, lower-cost AI models could actually expand overall computing demand through the 'Jevons Paradox.' Meanwhile, bottlenecks such as power supply, technical talent, and approval processes are more like 'speed bumps' slowing down construction, rather than 'brick walls' halting industry expansion.

Morgan Stanley forecasts that the combined capital expenditure of the U.S.'s five major tech giants will climb from nearly $800 billion in 2026 to about $1.2 trillion in 2027 and $1.4 trillion in 2028, providing highly visible revenue growth potential for assets such as GPUs, High Bandwidth Memory (HBM), Ethernet switch chips, optical communications, data center cooling, and power equipment.

Recovery Rally or a New Bull Market? The Market Remains Cautious

Despite the impressive rebound, most analysts remain cautious, viewing this more as a 'fundamentally supported recovery rally' rather than the beginning of a new leveraged bull market.

Data shows that while South Korea's KOSPI index surged nearly 18% in a single day, it still lags about 30% below its historical high. The Philadelphia Semiconductor Index fell over 20% in July, briefly retreating nearly 30% from its June peak.

The market widely believes that to confirm the true start of a new primary uptrend, several key indicators must be monitored: whether the semiconductor index can hold technical support during pullbacks, whether capital continues to flow in and drive upward earnings revisions, whether outflows from leveraged ETFs stabilize, and whether cloud providers can consistently prove that massive capital expenditure translates into actual cloud revenue and orders.

JPMorgan emphasizes that what truly supports the medium- to long-term bullish case for semiconductor stocks is the near-insatiable demand from data centers for advanced storage products such as high-bandwidth memory, server DRAM, and enterprise SSDs. Combined with disciplined supply-side management, both spot and contract prices retain upward potential.