Microsoft's latest earnings report could be a bearish signal for semiconductors such as storage and optical modules.

$Microsoft(MSFT.US)

The most noteworthy aspect of Microsoft's latest earnings report is not just the adjustment of capital expenditures from $190 billion to $175 billion, but the shifting logic behind AI infrastructure investment.

The market's past logic was:

Rising AI demand → Cloud providers frantically building data centers → Massive procurement of GPUs, CPUs, HBM, servers, optical modules, and power equipment.

Now, Microsoft might shift to:

Increased AI demand → Leasing mature data centers first → Directly installing GPUs inside.

However, if Microsoft reduces capital expenditures and adjusts its data center strategy, it will create short-term pressure on the AI infrastructure supply chain.

Sectors significantly affected include:

GPUs: If cloud providers slow down data center expansion, future GPU procurement schedules may be impacted.

Servers: Demand for AI servers relies on the capital expenditures of major cloud providers; a decline in CapEx will affect order expectations.

HBM and Storage: A slowdown in GPU deployment will impact expectations for high-bandwidth memory demand, putting pressure on the DRAM and HBM supply chains.

Optical Modules: Demand for 800G and 1.6T high-speed optical modules depends on AI data center expansion; slower construction will affect market expectations.

Data Center Infrastructure: This includes power, cooling, liquid cooling, transformers, and other segments, which will also be affected.

It is important to note that Microsoft is not abandoning AI investment, but rather changing its investment approach.

Past: Buying land and building data centers themselves.

Present: More reliance on leasing mature data centers and rapidly deploying GPUs.

The reason is that building large-scale AI data centers typically takes 2–4 years, whereas leasing mature facilities allows for faster compute deployment.

This implies that AI demand may still exist, but the supply chain is transitioning from "infinite expansion" to an "era of capital efficiency."

The market needs to observe in the future:

1. Whether Microsoft continues to reduce CapEx. 2. Whether Amazon, Google, and Meta are adjusting their capital expenditures in sync. 3. Whether NVIDIA GPU orders are slowing down. 4. Whether there are changes in HBM inventory and prices. 5. Whether growth in server and optical module orders declines.

If these signals appear simultaneously, it suggests that the AI infrastructure cycle may be moving from a high-speed expansion phase into an adjustment phase.

The core change is not the disappearance of AI, but the market shifting focus from "who invests more" to "who can achieve higher returns on investment."