Leopold Aschenbrenner's Situational Awareness fund made one of the more decisive institutional moves in the Bitcoin mining sector this past quarter, aggressively unwinding its exposure to publicly traded miners after a July transaction with Citadel. The fund — built around Aschenbrenner's thesis on artificial intelligence's trajectory and infrastructure demands — slashed its position in IREN by roughly 94% and made clean exits from Core Scientific, Riot Platforms, and Hive Digital in the third quarter. What remains is a residual $80.8 million spread across three miner positions — a shadow of what was once a meaningful stake in the sector.
The timing is worth examining closely. The Citadel sale in July appears to have served as a structural inflection point for the fund's posture toward Bitcoin mining equities. Rather than simply trimming around the edges, Situational Awareness chose the kind of wholesale repositioning that signals a fundamental rethink of the thesis — not a tactical hedge. When a fund named for strategic foresight walks away from an entire category of infrastructure equity at this scale, the market should pay attention to what that implies about the intersection of AI compute demand and Bitcoin mining economics.
The AI-Mining Overlap Is Being Stress-Tested
For the past two years, the narrative around Bitcoin miners pivoted heavily on their dual utility: facilities capable of pivoting between proof-of-work computation and high-performance AI inference or training workloads. Companies like IREN, Core Scientific, and Hive Digital leaned into this story aggressively, marketing their power infrastructure and data center footprints as assets that could serve AI hyperscalers just as readily as they served the Bitcoin network. Investors, particularly those with AI-forward mandates, found the overlap compelling enough to build meaningful positions.
Situational Awareness was among them. But the Q3 divestiture suggests the fund has concluded — at least provisionally — that the dual-use thesis either played out, is fully priced in, or no longer fits the fund's preferred risk profile at current valuations. A 94% reduction in IREN alone is not a rotation; it is a verdict. The same logic applies to the complete exits from Core Scientific, Riot, and Hive. These are not stocks you sell down to zero by accident or for liquidity management. These are intentional, declarative moves.
Citadel in the Picture
The July sale to Citadel adds another layer to this story. Citadel — Ken Griffin's multi-strategy hedge fund behemoth — is not an unsophisticated buyer. If Situational Awareness was offloading a portion of its miner book to Citadel in July, it raises the question of what each party saw in the transaction. Citadel may have been absorbing the position as a tactical trade, anticipating a short-term re-rating of miner equities, or as part of a broader basket play. For Aschenbrenner's fund, the sale provided an exit ramp before the more wholesale Q3 unwind.
The sequencing — Citadel sale in July, then systematic miner exits through Q3 — reads like a structured unwinding, not a panic liquidation. That discipline itself communicates something: Situational Awareness had a plan to exit, and it executed methodically. The $80.8 million that remains across three positions may represent strategic residual exposure, legacy holdings awaiting the right exit window, or positions tied to specific catalysts the fund still sees as live.
What Remains and What It Signals
The $80.8 million residual figure is not negligible — it represents a continued, if diminished, conviction in at least a subset of the miner universe. Without a full breakdown of which three stocks hold that capital, it is difficult to draw precise conclusions about where Situational Awareness sees remaining value. But the fact that the fund chose to retain anything after exits as aggressive as these suggests the divestiture is not a wholesale ideological break with the sector. It may instead reflect a sharper view of which mining operations have genuine data center infrastructure credentials versus those that are primarily leveraged Bitcoin price plays wearing an AI-infrastructure costume.
That distinction matters enormously right now. As power costs rise, as AI hyperscalers negotiate directly with utilities for dedicated capacity, and as Bitcoin's block subsidy continues its long-term decline, the miners best positioned to survive are those with real, contracted, diversified compute revenue — not those relying on narrative alone. Aschenbrenner, whose public writing on AI development timelines has earned him serious credibility in both tech and finance circles, appears to be applying that same infrastructure-first rigor to his portfolio construction.
The broader takeaway for the Bitcoin mining equity space is uncomfortable but necessary: institutional patience for the AI-mining convergence thesis has a limit, and one of the sector's more intellectually credible investors appears to have reached his. Whether that proves prescient or premature depends on how quickly the surviving miners can convert their power assets into contracted AI revenue at meaningful margins. The clock, it seems, is running.
Written by the editorial team — independent journalism powered by Bitcoin News.