Post

#财报观察员：微软云收入破千亿，Meta却指引拉胯——AI故事分化了？ Earnings reports on the same day, but the market reactions are completely different. Microsoft surged after hours, with Azure cloud revenue up 43% year-over-year, surpassing $100 billion for the first time annually. Copilot has 30 million paying users, and the order backlog stands at $678 billion. Enterprises are genuinely spending money. Meta fell more than 10% after hours; revenue grew 28% but profits declined, and free cash flow dropped from $8.5 billion to less than $800 million, the lowest in four years. Zuckerberg plans to continue ramping up, with capital expenditures of $130-145 billion next year. Both are talking about AI, but the fundamental logic is completely different. Microsoft is doing "AI + existing business." It hasn’t created a standalone new product but integrated AI into Azure cloud and Office, making users pay more. Enterprises already need Office, and now with the Copilot feature, renewals come naturally. Azure’s AI services work the same way: enterprise clients pay for compute power on demand. This model works because enterprise CIOs see efficiency gains and are willing to allocate budgets. Meta is doing "AI transforming algorithms." Its AI currently has no independent product and is mainly used to improve ad targeting accuracy and increase short video retention. The problem is this is hard to quantify—you can’t clearly say how much revenue growth is due to AI versus normal traffic growth. Advertisers don’t pay separately for AI; they only care about ROI. This makes Meta’s huge capital expenditures hard to justify with short-term revenue. Which model is more reliable? In the short term, Microsoft’s certainty is much higher. It has an established enterprise payment habit as a foundation, and AI is a value-add, not starting from scratch. Also, the $678 billion order backlog means most revenue for the next one to two years is already locked in, providing stronger risk resistance. Meta’s problem is "trust cost." The market doesn’t doubt AI can change advertising, but it doubts that throwing $130 billion in will yield equivalent returns. Especially with free cash flow dropping so sharply, investors lack patience to wait two or three years. Zuckerberg says "selling compute power for short-term profit is shortsighted," which is true, but capital market patience is limited. So how do I choose? My logic has two layers: Core position in Microsoft. The reason is simple: it has proven AI can convert into revenue, and it’s sustainable subscription revenue. Prioritize certainty; I don’t expect to get rich quick here, but I want stability. Small position in Meta, as buying a long-term lottery ticket. The premise is I won’t mind losing this money. Meta holds 3 billion daily active users, the largest attention pool in the world. If someday it successfully runs AI assistants or AI social, that space is indeed much bigger than Microsoft’s cloud services. But this may take 3-5 years to see results, with countless fluctuations in between; if you can’t hold on, don’t touch it. In short, Microsoft is making money now, Meta might make big money in the future. I’m betting on both, just with different proportions.