The AI Boom Faces a Leverage Problem After a Fund with 67% AI Investments Crashes

cryptonews.ruPublished on 2026-08-17Last updated on 2026-08-17

Abstract

Situational Awareness AI, a hedge fund founded by a former OpenAI researcher and led by Leopold Aschenbrenner, lost approximately 67% of its value in July 2026. The fund, which had generated a 439% return using roughly 4x leverage on concentrated bets in AI infrastructure and chip stocks, faced margin calls from major brokers as its positions fell 35-47%. Unable to meet the calls, the fund was forced to liquidate its entire public stock portfolio. The collapse highlights the risks of excessive leverage in volatile tech sectors, even when underlying investment theses—like strong demand for AI memory chips from companies like Micron—prove correct. Notably, the fund continues operating, having recently invested $400 million in a chip manufacturing startup. Global regulators, including the Bank of England, had warned of rising leverage in stock markets, particularly around highly valued AI companies, drawing parallels to past blow-ups like Archegos. While AI investment momentum remains strong, with trillions projected for infrastructure, the incident raises questions about how markets will manage risk. Institutional investors may shift toward longer-term, less leveraged structures, as seen in record fundraising for real assets like data centers in 2025. The key uncertainty is whether patient capital or highly leveraged holders will dominate during the next market downturn.

Situational Awareness AI, the hedge fund led by Leopold Aschenbrenner, lost approximately 67% of its value in July. This resulted from margin calls forcing the fund to sell its public portfolio. The collapse of such a fund raises questions for equity investors about the outlook for the second half of 2026. Despite accurately forecasting AI demand, the fund narrowly avoided catastrophe, making this event significant not just for one hedge fund manager.

Whether tech sector volatility will persist going forward is one of the key questions facing markets as the year draws to a close. The same companies that fueled stock index rallies can fall swiftly once leverage and liquidity decline, as July demonstrated.

Why a Winner with 439% Gains Lost Two-Thirds in a Month

Situational Awareness was founded by a former OpenAI researcher and had achieved 439% returns using roughly 4x leverage based on concentrated bets on AI infrastructure and chips. However, this leverage had its drawbacks. When its positions fell 35-47% in July, three major brokers, including Goldman Sachs and JPMorgan, demanded margin, which the fund could not meet, leading to the sale of its entire public stock portfolio by Citadel.

Reports confirmed the 67% drop in July and the sale of the fund's controlling stake in publicly held shares. According to Aschenbrenner, the company explained that the fund is prepared to 'continue the fight' and framed the decline in terms of risk, not investment strategy.

BREAKING: The full letter Leopold sent to his investors last night.

Leopold Aschenbrenner's fund fell 67% in July but is still up 80% year-to-date, and he announced he will continue investing in public equities. https://t.co/n3Swjmp8Z6 pic.twitter.com/jswe9ROkTo

— Leopold Stock Trac (Trac) July 31, 2026

'We welcome volatility. But it should never endanger the fund.' And the company has not left the industry, investing $400 million in Source Foundry, a lithography-focused chip startup valued at nearly $5 billion, specializing in the bottleneck of chip manufacturing, an area monopolized by ASML.

Demand Was Real, but Financing Was Lacking

The mistake was not the bet itself. According to documents, nearly half of the fund's US portfolio by the end of June was in companies like SanDisk and Micron, just as demand for memory chips was rising. Counterpoint Research found that enterprise solid-state drives accounted for 48 percent of total NAND sales globally in Q2 2026, during the transition of AI workloads from training to inference. Furthermore, TrendForce forecasts that NAND prices will rise 10-15 percent quarter-on-quarter, and DRAM chip prices will rise 13-18 percent quarterly.

Micron strengthened its market position with statistics, reporting an unprecedenteddentof $41.46 billion in revenue for its fiscal Q3 compared to $9.30 billion a year ago. However, this did not help the hedge fund. The Philadelphia Semiconductor Index fell nearly 30% from its June peak, and the aggressive position did not allow for waiting out the recovery suggested by demand data.

Central Banks Note Similar Leverage Buildup

The stated weakness wasdentby regulators even before the Situational Awareness case. A Bank of England Financial Stability Report published in July 2026 highlighteddentrisk of a 'significant increase in leverage on equity markets,' with particular focus on the stretched valuations of a small number of AI-linked companies. The Bank for International Settlements cited the 2021 collapse of investment fund Archegos and the 2022 UK liability-driven investment (LDI) stress to insist that non-bank funds take steps to avoid events like Aschenbrenner's. Expertsdentthat the greatest issues lie in leverage and concentration risks.

Essentially, the pattern remains the same: the right ideas, but the wrong form and method of implementation without any liquidity buffer.

Will Institutional AI Trading Persist?

A key question remains open for 2026 to address: how will the resulting gap be filled? JPMorgan suggests the tech sector may still retain a leverage effect that fueled its past rise and that levered ETFs, options, and margin accounts might still be in the stabilization phase. Such programming is expected to further shift the balance of demand for tech stocks between leveraged and unleveraged, with investors de-risking. According to PitchBook, real asset fundraising related to infrastructure, energy, and data centers reached a record $206.6 billion in 2025, with fund managers favoring long-duration structures and contracts,tracto leveraged directional investments.

AI investment volumes are not shrinking. Goldman Sachs estimates spending on computing power, data centers, and energy will total about $7.6 trillion. What changed in July is how much risk the market is willing to absorb and whether the next downturn will be met with patient capital or if highly leveraged equity holders will be forced to sell.

end-content

Trending Cryptos

Related Questions

QWhat happened to Leopold Aschenbrenner's hedge fund in July, and what was the primary cause?

AIn July, Leopold Aschenbrenner's hedge fund, Situational Awareness AI, lost approximately 67% of its value. This massive loss was primarily caused by margin calls from its brokers (including Goldman Sachs and JPMorgan) that the fund could not meet, forcing the sale of its entire public stock portfolio.

QWhat was the key investment strategy of the Situational Awareness fund, and why did it become a liability?

AThe fund's strategy involved using about four times leverage to make concentrated bets on AI infrastructure and chip stocks. This high leverage amplified returns, leading to 439% gains. However, it became a severe liability when its leveraged positions fell 35-47% in July, triggering the margin calls that wiped out the fund's public portfolio.

QAccording to the article, were the fund's core bets on semiconductor demand fundamentally wrong?

ANo, the core bets on semiconductor demand were not fundamentally wrong. The article states that demand for memory chips was actually rising, supported by data showing strong corporate SSD sales and projected price increases for NAND and DRAM chips. The fund's failure was due to its leveraged financial structure, not an incorrect assessment of market demand.

QWhat concerns have central bank regulators raised in relation to the incident with the Situational Awareness fund?

ACentral bank regulators, such as the Bank of England and the Bank for International Settlements, had previously warned about the risks of 'significant increases in leverage in stock markets,' particularly concerning overvalued AI-related companies. They highlighted the dangers of leverage and capital concentration, citing past examples like the Archegos collapse in 2021.

QLooking forward, what does the article suggest about the future of AI-related trading and investment?

AThe article suggests that investment in AI is not slowing down, with massive projected spending on computing and infrastructure. However, the key question is how the market will price and manage the associated risk. There may be a shift away from highly leveraged public market bets towards longer-term, less leveraged investments in tangible assets like data centers and energy, as seen in the record fundraising for real asset funds in 2025.

Related Reads

Trump-Backed Company World Liberty Financial Partners with Chinese AI Company of Limited Access to Provide Artificial Intelligence Services

According to a Reuters report, Trump-backed cryptocurrency company World Liberty Financial is linked to WorldClaw, a Hong Kong-based platform providing access to Chinese AI models that the U.S. government has flagged as security threats. WorldClaw's platform features AI models from Chinese companies like Alibaba, Baidu, Z.ai, DeepSeek, and Moonshot, some of which are designated by U.S. entities as tied to the Chinese military or subject to export restrictions. The connection is facilitated by World Liberty's dollar-pegged stablecoin, $USD1, which is accepted for payment on WorldClaw. The Trump family benefits from interest generated by the stablecoin's reserves, and token sales have earned them over $1.4 billion. A World Liberty executive previously advised WorldClaw on integrating $USD1. Security experts warn that using these Chinese AI models could expose users to surveillance or malicious code. WorldClaw claims over 10,000 users and processes 50 million queries daily, partly due to the lower cost of Chinese models. Both companies and the White House deny any conflict of interest, stating the platform helps American AI firms access international markets and that listing models does not imply endorsement. External critics, however, argue it is hypocritical to profit from Chinese AI tools while the U.S. government aims to counter the rise of Chinese AI.

cryptonews.ru22m ago

Trump-Backed Company World Liberty Financial Partners with Chinese AI Company of Limited Access to Provide Artificial Intelligence Services

cryptonews.ru22m ago

Moscow's Mining Ban Takes Effect: What Will Be the Consequences?

Mining has been banned in Moscow, the Moscow region, and parts of the Kursk region from August 15, with the prohibition set to last until December 31, 2032. The measure, requested by regional authorities, aims to prevent electricity shortages caused by energy-intensive mining equipment. This follows the legalization of cryptocurrency mining in Russia in November 2024, with restrictions now in place across 16 regions. Initially a blanket ban, an exception was introduced in March 2025 for miners using their own power generation, such as gas piston units or diesel generators, independent of the national grid. Major industry players had already begun shifting to on-site gas generation near gas fields to bypass grid limitations. While a trend emerged in 2025 to combine mining with AI computing infrastructure, this was complicated by authorities granting data centers (with a ban on mining inside them) preferential status as communication facilities. Experts state the new Moscow-area ban will significantly reshape Russia's mining geography but have a minimal global impact on Bitcoin's hash rate. The primary consequence will be a stronger industry shift to regions with energy surpluses, clear connection terms, and willingness to handle such loads. Moscow was already a contentious location due to high competition for power from data centers, industry, and urban infrastructure. For the Kursk region, the ban is also linked to infrastructure stability in border areas. The ban accelerates a market split. Large, legal operators will relocate equipment or seek sites with self-generation, while smaller or semi-legal miners operating in urban or industrial networks face higher costs and potential displacement into illegality. Industry participants note that post-2024 legalization had an unintended effect, with some major companies closing due to financial issues. Russian miners also face global pressures like low Bitcoin prices relative to mining costs, rising electricity prices, and the approaching 2028 halving, forcing some to shut down equipment.

cryptonews.ru25m ago

Moscow's Mining Ban Takes Effect: What Will Be the Consequences?

cryptonews.ru25m ago

Trading

Spot

Hot Articles

Discussions

Welcome to the HTX Community. Here, you can stay informed about the latest platform developments and gain access to professional market insights. Users' opinions on the price of AI (AI) are presented below.

活动图片