I think the market is treating Alibaba Group less like an e-commerce company now and more like a “China AI infrastructure + sovereign cloud” proxy. That is the real rerating driver.

The bullish case is not hard to understand:

Cloud revenue +38%

AI-related revenue still growing triple digits

Management saying AI products are already ~30% of external cloud revenue and could exceed 50% within a year

“No GPU sits idle” implies utilisation is extremely high, which matters because idle GPUs destroy ROIC in AI infrastructure businesses 

But the market is also glossing over something important: operational profitability is ugly right now. Adjusted net income collapsing toward near-zero while capex explodes tells you Alibaba is still in the “build first, monetise later” phase. 

The key question you raised is exactly the correct one: Can ¥380B+ capex convert into revenue fast enough before depreciation, pricing pressure, and competition catch up?

That depends on three things:

1. Whether Chinese enterprises genuinely shift workloads into AI/cloud at hyperscale.

2. Whether Alibaba can maintain pricing power instead of entering a margin-killing GPU price war.

3. Whether inference demand becomes durable rather than a temporary “everyone buys GPUs” frenzy.

I actually think Alibaba’s setup is structurally stronger than many Western AI names because it owns:

cloud infrastructure,

models (Qwen),

e-commerce traffic,

payments,

logistics,

enterprise distribution.

That vertical integration gives it more monetisation paths than pure-model companies.

But I would not blindly chase an 8% spike after a euphoric call. The stock is starting to trade on a forward narrative similar to early cloud-era Amazon or current NVIDIA expectations: “profits later, scale now”.

That works brilliantly until growth decelerates even slightly.

My view:

Long-term transformation? Plausible, yes.

Near-term valuation? Getting aggressive.

Best setup? Probably adding on pullbacks or post-hype consolidations rather than momentum-chasing after a vertical move.

If AI/cloud revenue truly compounds into 2027 while margins recover, today’s valuation may still look cheap in hindsight. But if capex outruns monetisation for another 2 to 3 years, the market could suddenly rediscover that cash flow still matters.