Same question, two sides

September's argument over Chinese versus American AI spending lands on Alibaba and Amazon. Both report strong compute demand. Shareholders still need it to cover the infrastructure bill.

Skip the capex beauty contest

Jefferies compared spending intensity, but headline capex misses the point. The better test is separating operating progress from cash committed against growth that has not arrived.

“distinguish operating progress from cash committed ahead of future growth”

Alibaba's cloud is working

June-quarter AI Cloud and Compute revenue rose 45% to RMB48.44 billion, with adjusted EBITA of RMB5.63 billion. The segment now folds in T-Head, so use the recast comparisons.

But the cash went out

Group capex hit RMB67.68 billion for the quarter and free cash flow ran negative RMB44.67 billion. Weaker returns elsewhere in Alibaba leave less room for error.

“Cloud growth alone cannot settle the value of the entire business.”

Amazon shows the same shape

AWS revenue grew 37% to $42.2 billion with $16.6 billion in segment operating income. Free cash flow was negative $7.6 billion over twelve months, blamed on AI property and equipment.

Watch cash, not restraint

Installation costs, competition and low utilization can all make returns worse than the growth rate implies. The evidence that matters next is cash generation rising alongside cloud revenue.

“Spending less is not automatically better; earning durable returns on the spending is what matters.”