Leopold Aschenbrenner is directing another $400 million into artificial intelligence infrastructure just days after his hedge fund, Situational Awareness, endured one of its most severe setbacks.

Aschenbrenner, a former OpenAI researcher who rose to prominence with his 2024 essay series arguing that AI infrastructure investment would define the decade, now manages capital around that same conviction.

According to a report by The Wall Street Journal on August 7, Situational Awareness has committed a total of $500 million to Source Foundry, with $400 million deployed this week alone. Source Foundry was founded in 2025 by Stanford researchers Abdulmalik Obaid and Joe Burg. Following backing from Sequoia Capital, the startup has been assigned a market valuation of approximately $5 billion.

The timing of the deal elevates its significance well beyond a standard venture investment.

July Losses and Portfolio Restructuring

In a letter to investors, Situational Awareness disclosed an unaudited net month-to-date loss of 67% for July. Despite this sharp decline, the fund's net year-to-date returns remained at 80%. Aschenbrenner explained that the fund had sold off a portion of its public portfolio to reduce leverage amid deteriorating liquidity conditions in the broader market.

The letter emphasized that the July drawdown did not alter Aschenbrenner's long-term conviction in AI. The fund will continue operating as a hybrid public-private vehicle. Going forward, its public portfolio will be managed on a fully paid-for basis, effectively eliminating all margin and liquidation risks.

Against this backdrop, the Source Foundry investment reads less like a doubling down on the leveraged strategy that contributed to July's losses and more like a sustained commitment to the physical infrastructure underpinning AI development.

Betting on the Semiconductor Manufacturing Bottleneck

Source Foundry is not simply another AI chip designer. The company aims to build machines, equipment, and software for semiconductor manufacturing, with a particular focus on lithography. Sequoia partner Stephanie Zhan described the company's mission as targeting the "tightest bottleneck" in semiconductor production, beginning with lithography.

This positions the startup in direct competition with ASML, whose extreme ultraviolet (EUV) lithography technology is indispensable for fabricating the world's most advanced semiconductors. ASML is the only company worldwide that produces EUV systems, making it one of the most concentrated chokepoints in the global technology supply chain—a position that has also made its equipment a focus of U.S. export controls aimed at restricting China's access to advanced chipmaking capability.

The technical hurdles facing Source Foundry are formidable. ASML's EUV systems rely on 13.5-nanometer light, generated by firing lasers at microscopic droplets of molten tin. The light travels through a vacuum and is steered by precision multilayer mirrors, as conventional lenses cannot effectively transmit EUV wavelengths.

No details have surfaced yet regarding how Source Foundry's approach differs from ASML's EUV architecture. What is clear, however, is that the startup is pursuing an alternative to the large, expensive, and highly complex equipment currently used in semiconductor production.

For the venture to succeed, Source Foundry's technology must extend beyond laboratory functionality. It would need to achieve levels of precision, throughput, and reliability sufficient for high-volume chip manufacturing, while also delivering improvements in production cost, equipment size, or deployment time.

Source Foundry's potential impact therefore extends well beyond a single AI chip company, as every advanced processor manufacturer depends on semiconductor manufacturing equipment.

Industry demand continues to climb. TrendForce forecasts that global foundry revenue will grow by 24.8% to $218.8 billion in 2026, driven by sustained demand for AI chips that keeps advanced node capacity consistently tight.

In other words, the events of July did not diminish the fundamental need for AI's physical infrastructure. Rather, they exposed the risks of financing that demand through leveraged positions in public markets.

Broader Implications for the Global AI Market

The significance of this investment may extend well beyond Situational Awareness.

As AI infrastructure scales, bottlenecks are migrating further upstream—from models and accelerators to memory, energy, advanced packaging, and ultimately to the machinery required to fabricate the chips themselves.

A joint analysis by the Semiconductor Industry Association and Deloitte found that semiconductors account for more than 95% of the total value in an AI server rack. Separately, AI data center infrastructure investment is projected to exceed $4 trillion by 2028.

This creates a notable tension for investors: equities tied to AI companies can decline even during a robust infrastructure buildout cycle.

Source Foundry represents a wager on what happens when surging demand collides with a more fundamental constraint—the machinery needed to manufacture sufficient volumes of advanced chips.

The investment also reinforces a lesson from Aschenbrenner's July losses. Correctly identifying AI's long-term growth trajectory does not guarantee survival through short-term market volatility. His 67% drawdown demonstrated how rapidly leverage and concentrated positions can become detrimental when market conditions shift abruptly and liquidity evaporates.

The $400 million commitment to Source Foundry signals that Aschenbrenner has not retreated from his AI thesis. Instead, he is redirecting capital toward a less visible and potentially pivotal segment of the technology stack.

The central question now is whether Source Foundry can translate that strategic insight into viable semiconductor processing technology. If it succeeds, Aschenbrenner's latest wager could carry even greater significance for AI than the public-markets position he lost, given that a breakthrough in chipmaking equipment would benefit not just a single AI accelerator manufacturer, but the entire semiconductor industry.