Stay connected with KayaToday—follow us on Instagram and Facebook for the latest news and reviews delivered straight to you.
As the U.S. Securities and Exchange Commission got back to work Monday morning, the financial world collectively held its breath for one document in particular, the Q1 2026 13F filing of Situational Awareness LP.
The fund, launched just over a year ago by 24-year-old former OpenAI researcher Leopold Aschenbrenner, has quickly gone from a niche, intellectual curiosity into one of the most intensively tracked portfolios on Wall Street.
The numbers alone are mind-boggling. The fund’s stated exposure to U.S. equities, which began with a seed of about $225 million, grew to $5.5 billion by the end of 2025 and now totals a jaw-dropping $13.67 billion through March 31, 2026.
But what holds the market in its grip is not the sheer velocity of capital creation, but the structural chutzpah of the capital allocation. Aschenbrenner is playing what can only be called the ultimate contrarian “barbell” trade of the artificial intelligence supercycle. On one side of the ledger he’s assembled a gigantic, multibillion-dollar long position in Bitcoin miners, alternative power suppliers and physical data center equipment. On the other hand, he has built $7.46 billion in highly leveraged derivative put options aggressively shorting the exact darlings of the AI boom: Nvidia, AMD, Broadcom, and big semiconductor exchange-traded funds.
To the casual observer, betting against the makers of the chips powering the AI revolution looks like financial suicide. But seen through the prism of macroeconomic restrictions, structural supply chain constraints and the blunt physical realities of energy security, the picture looks very different. The intelligence explosion is no longer constrained by algorithmic advances or software tuning. It is bound by megawatts, transmission lines, and the raw industrial physics of power generation .
The Paradigm Shift
If you want to grasp the logic behind a $13.6 billion shift from Silicon Valley darlings to heavy industries, you have to go back to June 2024. Aschenbrenner, for example, issued a 165-page manifesto called *Situational Awareness: The Decade Ahead*. The basic assumption was straightforward but terrifying: artificial general intelligence (AGI) is not a distant mid-century dream. It’s barreling down at us, probably before the end of the decade.
But Aschenbrenner correctly foresaw that the exponential growth of these frontier AI models would crash into the physical world, with a bang. To train the next generation of models you need gigawatt-scale computing clusters – huge facilities that use more electrical power than entire mid-sized cities. The current global power grid, built over decades for steady, predictable human use, is fundamentally unequipped to deal with the intense, concentrated energy demands of AGI data centers.
For the past three years the market has been fixated on a story of “intangibles”. Big bucks have gone into huge language models, application programming interfaces and cloud software suites. But Aschenbrenner knew early on that the real economic moat in the next decade will not be with those who create the finest code, but with those who control the raw physical inputs needed to run that code. The commerce has moved from software to concrete, from algorithms to electrons, from cloud architecture to heavy industrial supply chains.
The $7.46 Billion Chip Shortfall
Perhaps the most newsworthy finding in the latest 13F filing is the enormous size of bearish bets placed on the hardware giants. Situational Awareness LP reported a $2.04 billion put on the VanEck Semiconductor ETF and a $1.57 billion put that targeted Nvidia directly, as well as more than $1 billion in puts connected to Oracle and Broadcom.
This isn’t a general gamble against the usefulness of graphics processing units (GPUs) but an advanced macroeconomic hedge based on the truths of structural valuation and supply chain saturation. The market is basically pricing in 10 years of uninterrupted parabolic growth for GPU makers by early 2026. The premise is that demand will be greater than supply, and companies like Nvidia will be able to continue enjoying record, monopolistic profit margins, uninterrupted.
But Aschenbrenner’s options strategy paints a very different short-term picture. The physical infrastructure ( power generation and data center real estate ) is lagging behind chip production . That is a serious barrier . There’s no use deploying billions of dollars of high end GPUs if you don’t have the power grid capacity to turn them on. The fund is effectively shorting the overcapacity and the inevitable multiple-compression that happens when a high-growth sector suddenly crashes against a stiff, physical obstacle, by purchasing huge put options on the semiconductor layer. The chips are commoditizing . But the ability to run the chips is becoming the most rare asset on earth .
The Unlikely Guardians of AI Infrastructure
If shorting Nvidia is the protective hedge what is the aggressive long play? The answer lies in the most unlikely of places: bitcoin mining. Aschenbrenner has taken large positions in a number of Bitcoin miners including IREN, Core Scientific (CORZ), Riot Platforms (RIOT), CleanSpark (CLSK), Bitfarms (BITF) and Bitdeer (BTDR).
For years, traditional finance saw Bitcoin miners as speculative, unpredictable enterprises fully dependent on the wild price swings of digital gold. But Aschenbrenner sees them through the eyes of a commercial real estate and energy infrastructure expert. Bitcoin miners are essentially energy arbitrageurs. Over the past ten years they have combed the world, buying up enormous swathes of land, signing industrial-scale power purchase agreements and, critically, working their way through the maze of governmental permits needed to hook up to the main electricity grid.
Time is the currency of the race to create AGI. Today, an AI tech giant trying to develop a new gigawatt data center has to navigate years of zoning board meetings, environmental impact assessments, and grid connections queues that can extend well into the 2030s. But these assets are already in the hands of Bitcoin miners. They have the two scarcest resources of the AI age:
Grid-connected and allowed for real estate.
Deep discounts on long term power contracts.
Aschenbrenner is taking a great “shortcut” by buying into these mining companies. He is cutting through the years of red tape and getting the ability to operate, now. These miners are quickly upgrading their facilities from low-margin cryptographic hashing to high-margin, high-performance computing (HPC) for AI workloads. The shift from crypto miner to AI data center operator fundamentally rerates the underlying equities, converting a cyclical, volatile asset into a steady, high-yield utility stock.
And, there’s an interesting secondary macroeconomic effect at play here, too. As those miners repurpose their facilities to host AI workloads, they start generating revenue in hard U.S. dollars, not bitcoin. This flood of fiat liquidity greatly fortifies their business balance sheets and eliminates the fundamental necessity to perpetually liquidate their mined Bitcoin to pay operating costs. This removes the constant selling pressure from the market and structurally increases the Bitcoin floor price. Aschenbrenner has effectively gone long on energy scarcity, and at the same time created a supply-side shock in the crypto currency market.
Authentic Energy Independence
It’s not just stealing the existing electricity grid, it’s completely bypassing it. Besides the Bitcoin miners, Situational Awareness LP has amassed huge stakes in companies engaged in alternative, off-grid electricity generation. The biggest individual position in the portfolio is Bloom Energy (BE), a massive gamble of about a billion dollars.
Bloom Energy makes innovative solid-oxide fuel cells that can convert natural gas into energy on site, independent of the wider utility grid. This technology is the holy grail for an AI industry driven by unbroken uptime and huge power requirements. The current public grid is unstable, strained by the adoption of electric vehicles and sensitive to macroeconomic shocks and geopolitical energy embargoes.
Real energy security comes from having modular fuel cells right next to AI data centers. They safeguard their multi-billion dollar computing clusters from rolling blackouts, grid instability and the unpredictable supply chain shocks of global energy markets. This signifies a huge shift in how tech infrastructure is financed. We are seeing the privatisation of the electrical infrastructure, pushed by the insatiable demands of artificial intelligence.
The Macro Economics of AGI
The multi-billion dollar repositioning of Leopold Aschenbrenner should be a warning to the larger financial markets. The story of the AI revolution has been pulled, kicking and screaming, from the theoretical world of cloud computing and firmly into the gritty, capital-intensive world of macroeconomics, energy policy and supply chain logistics.
Investors who are still obsessed just with the computational prowess of the latest language models are ignoring the structural reality of the transaction. The bottleneck to the future is physical, not cerebral. It is measured in copper wire, industrial chillers, megawatts and interconnection queues.
Situational Awareness LP’s $13.6 billion bet is a masterpiece in recognizing the real restrictions of a paradigm-shifting technology. Aschenbrenner isn’t just riding the AI boom; he’s actively shaping its next economic chapter by shorting the consensus hardware trade heavily and aggressively amassing the basic physical infrastructure necessary to keep the lights on. As the decade passes, the real winners will be not the corporations that devise the most ingenious algorithms but the companies that own the electrons necessary to bring them to life.
Read Also: Bitcoin Reclaims $80,000+ After Its Best Month In A Year
