Cloud computing, that indispensable link in the AI industry chain.

After discussing storage, it's time to talk about cloud computing. After all, no matter how aggressively SanDisk and Micron ship, they ultimately have to sell to cloud giants like Google and Alibaba.

Many people worry about one issue—the larger the capital expenditure, the worse it is for cloud computing companies. You see Google and Alibaba spending tens or hundreds of billions of dollars each year on GPUs and building data centers. Upstream players like NVIDIA and SanDisk are making a fortune, while the cloud providers' own financial statements show a growing capital expenditure line. Is this business still viable?

I understand this concern, but I think the logic goes deeper. Because I've noticed another phenomenon: cloud computing can raise prices.

When computing power becomes a scarce resource, and when API call costs rise with electricity and hardware depreciation—cloud providers can fully pass the cost pressure downstream to the application side. Simply put, if you use AI, you have to pay for computing power; if you can't run the model, you have to keep buying cloud services. This logic is called "price increase" in other industries, but in cloud computing, it's "cost pass-through." There's only one prerequisite: the downstream side cannot do without you.

So, the question is: are consumers and businesses willing to pay for AI?

My view is relatively optimistic. Because paying to use AI on the production side has essentially become an irreversible trend. Customer service uses AI to respond to messages, designers use AI to generate images, and coders use AI to write code—once you use it, you can't go back to manual mode. Productivity is indeed improving; this isn't a bubble, it's real change.

Of course, the adoption rate is still very low. This also explains why the commercial returns from cloud computing don't seem explosive yet. Capital expenditure was spent upfront, but customers haven't fully caught up, creating a time lag. During this lag, cloud providers' profit margins are indeed under pressure.

But this situation will definitely improve. Because once AI's penetration rate passes a certain threshold, demand will rise exponentially. At that point, the pricing power in the hands of cloud computing will shift from a "potential weapon" to "real ammunition."

I've always been very bullish on Google and Alibaba, not just because of cloud computing. Google's search and advertising business, Alibaba's e-commerce and cloud ecosystem—they are more affected by consumption recovery in their respective countries. In other words, their stock price elasticity comes not only from AI but also from the recovery of the macroeconomy. This is a dual engine, but also a double-edged sword.

But what I want to say today is: cloud computing is the indispensable "middle layer" in the AI industry chain. It doesn't make money directly from price increases like storage, nor does it directly face consumers like applications. It connects the upstream and downstream—the value of upstream hardware can only be transmitted downstream through it, and downstream demand can only be fed back upstream through it.

This segment earns "infrastructure" money. The cycle is long, the returns are slow, but its position is rock-solid. As long as AI is still running and computing power is still being used, cloud computing will never be absent.

$Alphabet - C(GOOG.US) $BABA-W(09988.HK) $Sandisk(SNDK.US)