Reflections on the cooling AI boom: Why is the "Big Short" making a move now?

Last week, tech giants such as Amazon, Google, and Microsoft reported strong earnings, with capital expenditures remaining high. Coupled with the positive signals released by NVIDIA at the GTC conference, the AI sector saw an overall uptrend. However, as the week progressed, market sentiment cooled as the positive news was gradually digested.

Specifically, this adjustment was mainly influenced by the following factors: concerns over the high valuation of AI concept stocks, cautious attitudes expressed by some Wall Street institutions regarding the market outlook, and the phenomenon of "better-than-expected earnings but stock prices falling instead of rising" in certain companies. For example, Palantir's stock price fell by 7% after reporting better-than-expected Q3 earnings and raising its full-year guidance, reflecting profit-taking and fragile sentiment at high levels. Additionally, the ongoing U.S. government shutdown has also caused some disruption to market liquidity.

More notably, the recent disclosure of short positions in Palantir and NVIDIA by the "Big Short" prototype has further intensified discussions about whether the AI sector is in a bubble.

From a valuation perspective, the tech sector is indeed not cheap, but it hasn't detached from fundamental support. Tech giants generally have strong profitability and cash flow, differing from the previous market trends driven solely by "narratives." Cloud computing providers like Microsoft, Amazon, and Google have provided "hard constraints" for AI investment through their capital expenditures and order data. For instance, Microsoft Azure grew by about 40% last quarter, with commercial backlog reaching approximately $390 billion; Amazon AWS revenue grew 20% YoY, and Google Cloud grew 34%. These figures indicate that current AI investments still have solid business backing.

In the medium to long term, AI technology continues to evolve rapidly. The AI infrastructure race is heating up, with NVIDIA dominating computing power supply through its Blackwell chips and "AI factory" ecosystem, reaching a market cap of $5 trillion. Microsoft, Amazon, and Google have invested $80 billion, $125 billion, and $100 billion, respectively, to expand data centers. Azure, AWS, and TPU clusters support global model training; Meta has its own 1.3M+ GPU cluster and open-sourced Llama, while OpenAI partnered with SoftBank and Oracle to launch the $500 billion Stargate project, aiming for AGI-level computing power. Emerging players like xAI, CoreWeave, and Broadcom are also rising. Global AI infrastructure spending has exceeded $375 billion, and at least for now, corporate capital expenditures are driving AI development. Decode believes that, for some companies, the economic value of this will eventually match the scale of investment, given the substantial demand.

However, Burry's short position does have some merit.

Burry's logic is simple: the AI hype has created a new bubble. NVIDIA and PLTR are the "twin stars" of the AI narrative.

NVIDIA is the dominant player in AI chips, with its stock rising nearly 50% in 2025 and its market cap surpassing $5 trillion for the first time. Demand is indeed booming, with data centers being built frantically and GPU shortages persisting, but valuations clearly don't align with BV—the excess reflects expectations for NVIDIA's growth. This benefits from the Fed's rate cuts and the AI spending frenzy. If earnings fall short of expectations or supply chain/geopolitical risks emerge, the stock could crash.

Meanwhile, Palantir, the leader in AI data analytics software, has seen its stock surge over 100% this year, but its forward P/E is as high as 228x.

Burry is betting on "overheated growth expectations," similar to the subprime era when "everyone thought housing prices would rise forever."

In short, Burry believes that while AI investment has hit record levels, profits aren't keeping up with the hype: cloud growth is slowing, data centers resemble "Subprime 2.0," and everyone is betting on an "AI perpetual motion machine." If the Fed doesn't continue cutting rates or the economy fails to achieve a soft landing, these "AI darlings" will be the first to suffer.

Looking ahead, a few key developments warrant attention. In the short term, market sentiment may continue to be influenced by NVIDIA's earnings. Dazi will report earnings on November 19, with the market focusing on Blackwell chip supply progress and gross margin changes. If earnings or guidance disappoint, sector volatility could follow.

Additionally, the capital expenditure trends of cloud providers will be crucial in assessing the authenticity of AI demand. Microsoft, Amazon, Google, and others are maintaining high investments, and future marginal changes in their capex plans will need monitoring.

While the market faces short-term adjustments, these are largely driven by sentiment factors like the government shutdown and bubble theories—with no significant deterioration in fundamentals yet. AI giants' valuations remain within acceptable ranges under current interest rate conditions, and the 2026 earnings outlook provides some support. Burry may have his own plans, but there's no need for excessive panic in the near term.

Therefore, at current levels, Decode prefers to stay calm and avoid overreacting to short-term fluctuations. In the medium to long term, the actual implementation of AI technology and productivity improvements remain the main themes, with the key being to track corporate profitability and cash flow performance. For investors focused on the AI sector, it may be wise to take a long-term view, focusing on technological progress and commercialization timelines rather than short-term sentiment.

$NVIDIA(NVDA.US) $Palantir Tech(PLTR.US) $Amazon(AMZN.US)