357 billion US dollars evaporated! Microsoft Crash: The Death Kells of the AI Bubble Are Officially Sounding?

After announcing a financial report that exceeded market expectations, Microsoft suffered the most severe market sell-off of the year.

On Thursday, software giant Microsoft crashed 7% in premarket trading, wiping out $357 billion in market capitalization in a single day — a figure worth the disappearance of an entire Coca-Cola company from the face of the earth. And on the same day, Microsoft just handed over a near-perfect report card: profits rose by nearly a quarter and revenue hit a record high.

! [image.png] (https://img-cdn.gateio.im/social/moments-2b02ca5507e03737914135597fc16f48 “1769727665204057.png”)

Profits are rising, but stock prices are collapsing. Behind Wall Street’s vote with its feet, there is a tacit fear: the more the AI account is counted, the more shocking it becomes, and the story of growth has come to an end? When all the giants are crowded on the same track, who can really make money?

Beautiful “face” and the “inside” of cracks

From the first line of the financial statements, Microsoft’s performance is strong:

• Net Profit: Adjusted net profit increased 23% to $30.9 billion, higher than analysts’ expectations of $28.9 billion.
• Revenue: Up 17% to $81.3 billion, also beating expectations of $80.3 billion.
• Cloud business: Quarterly revenue exceeded $50 billion for the first time.

! [image.png] (https://img-cdn.gateio.im/social/moments-3e6fef90be017e0ad0f2eb4dc4b625b4 “1769727579296931.png”)

However, the market’s attention quickly focused on two details: the pace of Azure’s growth and the pace of capital expenditure expansion.

According to the financial report, Azure’s year-on-year growth rate was 38%, which is still strong but down one percentage point from 39% in the previous quarter.

Against the backdrop of historically high valuations, this 1% slowdown is seen as a signal of “growth peaking”. Barclays analysts bluntly pointed out: "Even with the overall digital health, buy-side investors clearly want to see more. ”

Microsoft has become a “hardware worker”?

In this round of AI gold rush, although Microsoft is the frontrunner, it is more like a “high-end foundry”.

Behind the astronomical investment is the extremely cruel pressure of hardware premium. According to TrendForce’s latest industry report as of January 2026, HBM (high-bandwidth memory), which is the core supporting equipment of the NVIDIA B200 series, is experiencing an unprecedented “capacity hijacking”.

Data monitoring shows that the average selling price of HBM chips has bucked the trend by about 30% in the past two quarters as HBM orders from Micron and SK hynix have been generally scheduled until early 2027. For Microsoft, this is tantamount to a kind of “structural blackmail”. To ensure that Azure AI’s computing power is not left behind, Microsoft must accept a premium of thousands of dollars per chip.

! [image.png] (https://img-cdn.gateio.im/webp-social/moments-cfc6c9f13c8c8efa79195909544ea753.webp “1769727887899287.png”)

Horizontal comparison of core data of cloud giants (Q4 2025 – Q1 2026)

What is this concept?

Most of the tens of billions of dollars Microsoft invests every quarter go to upstream hardware vendors. This means that the money earned by Microsoft was transferred to Nvidia to buy graphics cards and Micron to buy memory before it was heated on the books. Although Microsoft is also developing its own Maia chip, it is still highly dependent on external procurement so far. The result was brutal: the gross profit margin of Microsoft’s cloud business has slipped from a high of over 70% to around 67%.

On the other hand, Amazon AWS’s operating profit margin remains stable at 38% by reducing its dependence on expensive hardware with its self-developed chips (Trainium series) laid out in the early years. Although Meta is also investing heavily, its stock price has risen by 10% due to “seeing receipts” because AI has directly increased advertising conversion rates. In contrast, Microsoft is more like “working” for hardware manufacturers.

This “bleeding” investment not only failed to satisfy the appetite of the market, but induced Microsoft’s unique computing power “internal friction”. Due to limited supply, Microsoft has to face a cruel balance: should it rent out the top computing power to external cloud customers to earn instant profits, or leave it to its own Copilot to win the ecological future? Microsoft chose the latter. Although this strategy stabilizes the product experience, it has severely diluted Azure’s profitability as a cloud-only platform in the short term.

Concentration anxiety: OpenAI’s “single point dependency” crisis

In this financial report, Microsoft disclosed a shocking data for the first time: about 45% of the book value of its $625 billion future cloud contract came from OpenAI.

This means that Microsoft’s cloud growth is highly tied to a startup. Although CFO Amy Hood emphasized that there is still $350 billion from customers in other industries, investors are still worried that if OpenAI stalls in the competition or switches to autonomous hardware in the future, Microsoft’s costly system will face serious “idle risk”.

The dissolution of the moat: open source and low-cost dimensionality reduction strikes

In addition, the strong binding with OpenAI is facing a dimensionality reduction blow of a “cost-effective revolution”.

With the rise of low-cost or open-source models such as China’s DeepSeek, the “price war” in the AI market has begun. Microsoft’s high-premium Copilot subscription fee model is facing challenges when enterprise-level customers discover that an open-source model of a few cents can solve 90% of problems.

The uncertainty of this business model makes Microsoft’s high price-to-earnings (P/E) ratio seem shaky. If Microsoft can’t prove that its high computing power costs translate into equally high premium revenue, then the moat it has built may be quietly being flattened by the open source wave.

In the face of the downward trend in stock prices, Nadella remains unwavering. He touted his vision of “full-stack AI” on analyst calls: “When you think about our capital expenditures, don’t just think about Azure, think of Copilot.” We don’t want to maximize just one business, we allocate capacity to build the best portfolio of assets. ”

Epilogue

Despite the panic sell-off in the market, the giant is stabilizing its position through a series of complex capital operations.

! [image.png] (https://img-cdn.gateio.im/social/moments-c9ef6740d0c28df04cec508587e2329c “1769730545933508.png”)

Microsoft disclosed one during the quarter Accounting income of $7.6 billion, all thanks to its early investment in OpenAI. As OpenAI restructured from a nonprofit to a traditional for-profit in October, its balance sheet swelled sharply with multiple rounds of massive funding. Currently, Microsoft holds the AI leader 27% shares. As OpenAI seeks to surpass $750 billion The initial $14 billion invested by Microsoft has been exchanged for a staggering book return.

This left-footed right-footed “ecological flywheel” is becoming more and more complex: competitor Anthropic has just promised to buy $30 billion worth of Azure computing power in the future, and Microsoft plans to inject $5 billion into it. In this potential deal, the startup’s valuation has been pushed up to $350 billion.

In summary, Microsoft’s evaporated market value of $357 billion is a correction to its “capital-heavy and slow monetization” model. Despite the extremely generous paper gain on the books, what Wall Street really cares about is not how much valuation premium Microsoft earns as a “venture capital institution”, but whether its core cloud business can really recoup real money from global enterprises under the encroachment of hardware costs.

At this moment, the AI industry is like a high-speed train: once the infrastructure is rolled out, it is difficult to stop, and whether it can gradually achieve a closed loop of business while maintaining speed will determine the next stage of market pricing logic.

Author: Bootly