A news item about computing power from Meta punctured the one-sided narrative of momentum trading in AI hardware.

According to Mars Finance, Meta's stock price surged on Wednesday, July 2nd, but this unexpectedly put the AI hardware market under pressure. The market had initially expected a relatively quiet start to July. The second quarter saw a strong performance in the US stock market, with the S&P 500 recording one of its best quarters since the resurgence of the COVID-19 pandemic in 2020. However, before the US market opened, news that Meta might release or sell "excess computing power" suddenly changed the market narrative. This news was positive for Meta itself. The market interpreted it as the company shifting from continuously increasing capital expenditures to emphasizing financial discipline and free cash flow. Meta's stock price subsequently surged, with the article stating that it rose by about 10% in a single day, one of its best single-day performances this year. But for the AI hardware chain, this is a different story. One of the most crowded trades in the market over the past few months has been betting on cloud vendors continuing to expand their computing power, storage, and data center capital expenditures. If Meta begins to release excess computing power, investors will naturally ask: Is the demand for AI computing power really as unlimited as previously expected? Is cloud vendors still only revising their capital expenditures upwards, with no downward revisions? UBS trader Christina Dwyer stated that the Meta event pushed the market narrative towards "stronger financial discipline," while alleviating concerns about continuously rising capital expenditures. This benefits platform technology companies whose valuations are already nearing low levels, but weakens the "long-term computing power shortage" logic that previously supported neocloud, semiconductor, storage, and AI supply chain stocks. The market reacted quickly. The BofA Neocloud Basket fell significantly, storage and momentum stocks were impacted, and previously surging stocks like SanDisk and Micron were sold off. Micron was particularly considered a key watch: it had held above its 20-day moving average since April, and a break below it could open up room for a pullback to the 50-day moving average, implying a potential downside risk of about 20%. This correction also quickly evolved into a momentum trading clearing. Jonathan Krinsky of BTIG pointed out that the Bloomberg Mag7 index, relative to the Philadelphia Semiconductor Index (SOX), saw its largest single-day rebound since 2015. In other words, funds are flowing back from chip, storage, and high-beta AI hardware stocks to large platform technology stocks. Goldman Sachs' high-beta momentum basket fell about 9% in a single day, and the long-short high-beta momentum portfolio fell about 10%, nearing its worst performance since the 2020 vaccine news shock. The Meta event reminded investors that cloud vendors who actually bear capital expenditures also calculate returns. Once the certainty of upward revisions to capital expenditures decreases, the segments with the largest gains and most crowded valuations in the AI hardware chain will be the first to come under pressure. However, funds have not completely left the AI theme. Software stocks have outperformed semiconductors, and Bitcoin has also rebounded due to funds withdrawing from AI/storage momentum trading. Another beneficiary is other AI bottleneck assets such as capacitors, indicating that the market is still looking for scarce links in computing infrastructure, but is no longer indiscriminately chasing storage and chip momentum stocks. Another risk that the market needs to pay attention to is the current poor liquidity. Goldman Sachs trading desk stated that liquidity at the top of the S&P E-mini market fell by 33% month-on-month in June, while US stock trading volume hit its highest level since 2026. This means that while the market appears active, its actual absorption capacity is weakening; once large sell orders appear, prices are more prone to sharp fluctuations. The true meaning of the Meta event may not be Meta itself, but rather that it hit the most sensitive spot in the AI market: whether capital expenditure will continue to grow without limit. AI demand has not disappeared as a result, but the market has begun to distinguish between two types of companies: one is platform companies that can recoup their stable investments in computing power, and the other is hardware and storage suppliers that have already fully reflected the expectation of computing power shortages.