In this round of earnings reports, the biggest change in AI infrastructure isn't how much money was spent.

It's that the people paying have changed.

Previously, the market worried about:

Data centers being built more and more,

but in the end, only companies like OpenAI and Anthropic, which are building large models, were willing to keep paying.

No matter how busy the construction site is,

if too few people are paying,

and the money just circulates among a few companies,

this loop isn't stable enough.

But this quarter, more enterprises have started to truly enter the market.

Microsoft stated that nearly 90% of Microsoft Cloud's revenue over the past year came from customers outside these large model companies.

More intuitively, the number of paid seats for Microsoft 365 Copilot has exceeded 30 million.

The new paid seats added this quarter are more than double those of the previous quarter.

Enterprises are not just buying computing power in the cloud,

they are also integrating AI into their daily workflows, paying by seat and by actual usage.

Microsoft's Cowork (another AI tool) also has thousands of enterprises paying in this manner.

AWS corroborates this from another angle:

AWS's revenue growth accelerated from 28% to 37%.

At this current pace, AWS's AI business and chip business will each exceed $25 billion annually.

What I see is not just faster cloud revenue growth.

Enterprises first calculate some value from AI, and are willing to continue buying seats, usage, and cloud services. Cloud providers see increased revenue and orders, giving them the confidence to continue investing in data centers, chips, and power. The next step is to see if stronger, cheaper AI can help enterprises calculate even greater value.

This is where the positive cycle begins to form.

Of course, it hasn't fully run its course yet.

If enterprises feel it's not worth it, and renewal and expansion slow down, cloud providers' revenue, investments, and upstream orders will come under pressure.

But the question of "who ultimately pays" is now much clearer than last quarter.

Who ultimately pays determines how long this round of AI infrastructure can sustain itself.

Do you think that when enterprises buy AI,

the easiest benefit to realize right now is cost reduction or revenue increase? 🤔

$Amazon(AMZN.US)$Microsoft(MSFT.US)$Meta Platforms(META.US)$Alphabet - C(GOOG.US)