원문 보기
Michael Beugry became a Wall Street legend with just one guess in 2008. But over the past three years, it has burned through tens of millions of dollars trying to catch the peak of the AI revolution. The quantitative results of his macroeconomic “ bearish” bets are brutal and undeniable. In 2023, Bugry invested $1.6 billion in notional put options on the S&P 500 and Nasdaq , but was forced to exit the position with a loss of about 40% due to the rise of the technology sector.
Shortly after, he bought put options on the semiconductor ETF SOXX with a strike price of $490, just before the index rose another 16%, completely evaporating his option premiums. Then there was his high-profile $1.1 billion short bet against Nvidia and Palantir, where he placed Nvidia put options with a strike of $110, only to see the semiconductor giant hit a historic market cap, leaving his derivatives deeply unprofitable.
Yet, despite an endless string of failed bets, the icon of the " Great Short Sale" refuses to back down. In his latest reports, short positions and put options account for over 21% of the total portfolio, aggressively targeting five major tickers in the technology and industrial chains. Nvidia, Oracle, Palantir, NVR and Caterpillar.
But if you carefully examine his trading mechanics, you can see the quantitative panic hidden beneath the surface. Bugry just allocated nearly 4% of his portfolio to Nvidia December call options with strike prices in the $250 range . Paying small premiums for the contract, he openly called the deal “ growth insurance,” bluntly acknowledging that post- report volatility spikes are literally a coin flip that can instantly liquidate an unhedged short position.
This creates a bizarre evaluation trap for the contrarian master. On paper, Bugry admits that Nvidia looks heavily undervalued with a forward P/E of around 23.5, generating over $96 billion in quarterly revenue with a net profit margin of 62%. However, he insists that Big Tech's gigantic capital spending on equipment will inevitably lead to a serious oversaturation of the market, causing a sudden and devastating profit shock.
So, is Beugry’s systematic call-option hedging a masterclass in risk management, or proof that hyper-growth technologies will crush him again? Let's break down his five short goals one by one to see where the data is really pointing. The first is Nvidia, the main anchor of Bugry's " bearish" thesis , where he argues that the chipmaker is using dangerous crowdfunding to artificially inflate demand for GPUs.
Bugri argues that guaranteeing customer debt and supporting the residual value of up to $500 billion in AI infrastructure is classic vendor financing that hides the inevitable collapse in demand for computing power. But the raw quantitative data suggests that Bugri is fighting an economic giant , not a house of cards.
Nvidia doesn't just sell silicon chips. It is behaving like an AI state, strategically subsidizing its digital ecosystem to ensure the dominance of the CUDA platform for decades to come. Wall Street’s concerns ignore the fact that Nvidia is partnering with Apollo, BlackRock, Blackstone, KKR, Brookfield, and Goldman Sachs to provide up to 25% residual value support on $500 billion in infrastructure, as well as providing $105 billion in guarantees on OpenAI’s data center leases.
This is creating artificial demand. This is the rational development of an ecosystem with asymmetric information advantages. Look at the metrics that Barry is trying to "short ". Nvidia boasts a stunning net margin of 62%, while its implied price-to- earnings (P/E) ratio has fallen from 43.7 to 20.9-23.5, making the company cheaper than the Nasdaq average of 28.4 despite forecast revenue growth of 61.96% through fiscal 2028, which is expected to reach $665 billion.
Moreover, the demand is expanding far beyond advanced laboratories. Nvidia's AI Cloud, Industrial and Enterprise Solutions segment grew 138% year-over-year to $40.3 billion , driving nearly 59% of total data center revenue growth sequentially . The return on investment is already becoming real, as seen in the 14% increase in Meta ad impressions and the 12% increase in average ad prices, which is directly driven by AI recommendation engines.
With $279 billion locked in procurement commitments to fulfill orders, Barry's "short" ignores fundamental market reality. Nvidia is not creating artificial demand, it is experiencing an acute shortage of supply. Next on Barry's list of targets is Palantir, where he bought March 2027 put options with a strike price of low to mid- $100, betting on a massive collapse from a peak of over $200 to the $100 mark.
But betting against Palantir's operating leverage turns out to be a statistically fatal gamble. While Barry focuses on high valuation multiples, the company's internal fundamental engine is growing at an unprecedented pace. Palantir's quarterly revenue rose from $1.18 billion to $1.94 billion in just three quarters, driven by triple-digit growth in its U.S.
commercial division, where customer conversion rates remain well above 100%. Even more impressive is its hyper-profitability. GAAP net profit margin increased from less than 10% at the end of 2024 to a massive 55% in the second quarter of 2026, pushing GAAP operating margin to 47% and bringing the “Rule of 40” to a global level of 155%.
As earnings grow faster than the stock price, Palantir's valuation multiples are actually contracting even when the stock is trading near its highs, consistently forcing Wall Street analysts to raise revenue forecasts after each successful quarterly report. Berry is counting on retail price volatility to drive Palantir down to its $100 strike price, but shorting an enterprise software monopolist that turns demand for enterprise AI into 55 % of GAAP net profit does n't look like searching for a price bubble, but rather like trying to stop a runaway train .
Third on Berry's list is Oracle, where he is betting that the company's aggressive expansion of AI data centers will collapse under a mountain of debt. Berry considers Oracle's staggering $638 billion in outstanding liabilities an accounting illusion, arguing that a massive $70 billion capital expenditure through 2027 will destroy the balance sheet as consumer demand evaporates.
But the real data refutes his bearish forecast at every level. Oracle ended fiscal 2026 with revenue of $67.4 billion , operating income of $28.9 billion (non- GAAP), and a 54% jump in operating cash flow to $32 billion . Within the cloud division, infrastructure revenue accelerated 93% year-over-year to $5.8 billion .
Meanwhile, revenue from CPU and GPU infrastructure grew by 119% to $4.8 billion. Oracle added 1.2 gigawatts of data center capacity in 2026 and is adding another 1 gigawatt in the first quarter of 2027 alone. Oracle doesn't build empty data centers on empty hopes. A staggering 98% of available AI capacity has already been contracted, and actual infrastructure utilization is 97.5%.
Moreover, the balance sheet risk that Berry fears is largely borne by customers themselves, as most of the $85 billion in new quarterly order book came through direct prepayments and a “bring your own equipment” structure, where customers supply their own expensive accelerators. With a forward P/E ratio of just 18.2, 21% below the broader tech sector, despite forecast revenue growth of 34 % to nearly $90 billion in 2027, Oracle's fundamentals support a $185 price target based on expected EPS of $8.05, making Berry's short bet increasingly detached from operating reality.
Fourth on Berry's list is Nebius Group, a fast-growing neo-cloud provider where he is betting against the financial sustainability of high-tech infrastructure projects. Berry likely views Nebius as a speculative trap with high share dilution, pointing to a $5.75 billion convertible bond offering and active share dilution to fund a massive capacity expansion.
But again, the real numbers demonstrate an explosive fundamental engine. In the second quarter of 2026, Nebius delivered a stunning 454% year-over-year revenue jump to $582.3 million, while the annual growth rate soared by nearly 600% to $3.0 billion. Far from being a financially insolvent “black hole,” Nebius generated $236.2 million in adjusted EBITDA, delivering an elite EBITDA margin of 41% and a gross profit margin of 74%.
Importantly, Nvidia owns a direct 9.3% equity stake in Nebius after a $2 billion direct investment, guaranteeing the distribution of its GPUs and effectively eliminating the underlying execution risk for Neo clouds. Nebius received a loan at SOFR plus 2.50% , while spot GPU rental prices remain at 52% of the asking rates , demonstrating the structural supply constraints in the market.
With new data center deployments generating over $20 million per megawatt of annual contract value with a fast payback period of 22 months, Nebius achieves a return on invested capital of over 30%. Add to that non- core assets including a 28% stake in ClickHouse valued at $15 billion, Avride's $6 billion fleet of autonomous vehicles, and Toluca Arena, and you see Berry playing short on a company whose underlying infrastructure and assets could easily deliver 300% value growth over the next 5 years.
Fifth on Berry's list is Caterpillar, where his short position suggests a broader macroeconomic bet against the physical infrastructure powering the AI revolution. Berry sees Caterpillar as a bloated, late-cycle industrial company whose premium forward P/E of 33.5 times will collapse once hyperscaler data center construction slows.
But operational data shows an industrial giant operating at full capacity, quietly becoming a significant direct beneficiary of AI’s electricity needs. In its most recent quarter, Caterpillar posted revenue growth of 24 % and a stunning 73% increase in diluted earnings per share to $8.17, while operating profit jumped 50%.
The secret engine of this acceleration is power generation, where revenue grew 72% year-over-year as Big Tech competes for powerful power generators to support the operation of huge data centers. With a whopping $72 billion in order backlog and a net margin of 13%, nearly double the industry average, Caterpillar is using its record cash flow to aggressively repurchase $1.5 billion in shares per quarter.
While Berry is betting on Caterpillar shares to fall , suggesting that spending on AI infrastructure is an unsustainable surge, Caterpillar's record order book and dominant position in global power generation suggest that the physical deployment of AI is just beginning.
Comments 0