The debate about a bubble resurfaces.

Xie Yingjie, Investor Network

Since the fourth quarter, the artificial intelligence (AI) sector in the U.S. stock market has cooled off, transitioning from a state of exuberance to one of cautious observation. Oracle, a tech behemoth with a market capitalization nearing $300 billion, has become the latest focus for short sellers.

Recalling the market trends, Oracle's stock price soared to an all-time high in September, achieving record-breaking gains before embarking on a downward trajectory. By December, Oracle was hit by a barrage of negative news, which not only weighed on its stock price but also intensified discussions about the divergence in prospects and valuation bubbles within the AI industry. As of the close on December 17th, Oracle's stock price was at $178.45, marking a 5.4% drop for the day and a cumulative decline of 48.5% over the previous three months.

As one of the key drivers behind the U.S. stock market's bull run over the past two years, the future moves of the AI sector remain a crucial factor in determining the direction of the U.S. stock market next year.

Oracle Sparks a Sell-Off in the U.S. Stock Market

On the news front, the Financial Times reported that Blue Owl Capital, a significant data center partner of Oracle, had scrapped plans to finance Oracle's $10 billion data center project in Michigan. This news triggered a sharp decline in the once-hot AI concept stock. However, Oracle later refuted the report, asserting that the project was progressing smoothly.

Investors are expressing their concerns through their financial decisions, questioning not the demand for AI itself but when the heavy-asset closed loop of 'borrowing-expansion-delivery-return' will yield cash flows commensurate with the massive investments made.

Jacob Bourne, an analyst at Emarketer, commented, "Oracle itself is also under increasing scrutiny, with a focus on its debt-fueled data center construction and the concentration risk of its business amid uncertain prospects for AI spending. The revenue shortfall may exacerbate investor concerns, already cautious, about its deal with OpenAI and substantial investments in AI."

Over the past two years, the AI narrative has been fueled by OpenAI, propelled Nvidia to prominence, and swiftly followed by Microsoft, Google, and Amazon, leading to an exponential expansion in capital expenditure races. Although a latecomer to the 'cloud-native' scene, Oracle has successfully entered the 'computing infrastructure' concept stocks with $523 billion in unfulfilled contracts signed with OpenAI, Meta, TikTok, and others, driving its stock price upward.

However, when the earnings season truly tested the speed of monetization, the market discovered that, on one hand, GPU delivery cycles lengthened, and data center power and land costs soared, continuously driving up capital expenditures. On the other hand, the price war for large models intensified, while cloud computing rents declined, squeezing profits from both ends. Oracle's input-output ratio deteriorated significantly faster than its revenue growth.

What makes investors even more uneasy is the financing structure. To support its $50 billion infrastructure blueprint, in the second quarter of fiscal year 2026 (ending December 2025), the company's quarterly free cash flow turned negative at -$10 billion, the first such occurrence since 1992. Analysts anticipate its cash flow to remain negative until 2029. This suggests that despite increasing debt, cash reserves may not have grown in tandem and could even be depleted to cover capital expenditure shortfalls.

Moreover, Oracle's capital expenditures surged in 2025, reaching $12 billion in the second quarter, far surpassing expectations. It plans to boost its full-year capital expenditures for fiscal year 2026 by approximately $15 billion to $50 billion, primarily for AI-related infrastructure. Morgan Stanley predicts that its adjusted debt could reach around $300 billion by 2028.

Market Discussions on Sectoral Bubbles

In 2025, artificial intelligence emerged as one of the top-performing sectors. Since the fourth quarter, the AI sector has undergone a correction, sparking debates about sectoral bubbles.

The industry grapples with the dilemma of investment scale far outstripping short-term returns. OpenAI alone plans to invest $1.4 trillion over the next few years. Tech giants like Alphabet and Microsoft also intend to spend over $400 billion on AI-related capital expenditures in the next 12 months, mostly for data center construction, but related revenues are far from covering costs. This 'burning money for growth' model heavily relies on external funding, and any disruption in financing could trigger a chain reaction.

Massive capital expenditures have also raised concerns about return on investment (ROI). Although AI is widely seen as one of the most promising technological directions for the future, its commercialization path remains unclear, and its profit model is still in the exploratory stage. There is considerable uncertainty regarding how much profit it can generate in the future.

However, the 'bubble camp' does not hold sway over the discourse. Morgan Stanley highlighted in a concurrent research report that while AI capital expenditures are high, they are far from reaching an industry-wide surplus. Demand is just starting to be unleashed, and leading companies' balance sheets are stronger than during the telecom infrastructure boom before the millennium. "The market is confusing short-term profit friction with long-term value destruction."

Morgan Stanley pointed out that artificial intelligence is in the early stages of large-scale investment, with global related spending projected to approach $3 trillion by 2028. This wave of spending marks the beginning of a multi-year AI capital expansion. Hyperscale cloud service providers are the primary drivers of investment growth. Currently, industry procurement commitments have reached $330 billion, with lease agreements totaling about $340 billion, indicating the scale and intensity of capital investment. Supported by standard software profit margins, AI software revenue is expected to reach $1.1 trillion by 2028, suggesting significant return potential for invested capital.

HSBC believes that the AI investment cycle will continue to bolster corporate earnings. Citibank anticipates that the benefits of AI investment will keep being realized, but market attention will shift from AI-enabled companies to application companies.

Looking ahead, the AI industry may enter a 'validation phase.'

According to IDC data, the global AI server market is expected to reach $125.1 billion in 2024, growing to $158.7 billion in 2025, and hitting $222.7 billion by 2028. The compound annual growth rate from 2024 to 2028 is projected to reach 15.5%, with the proportion of generative AI servers increasing from 29.6% in 2025 to 37.7% in 2028.

Upstream chip manufacturers are poised to see increased production of Nvidia's B100 and AMD's MI400, narrowing the supply gap. Midstream cloud providers' capital expenditure intensity is likely to peak in the second half of the year. Downstream model competition will also shift from parameter races to cost, security, and compliance. Whether high investments can be translated into sustainable cash flows during this window will determine the direction of the 'bubble.' (Produced by Sino Wealth Management) ■

Source: Investor Network