Microsoft and Meta post strong earnings but take opposite reactions from markets
Jan 29, 2026
Key Points
- Microsoft beat Q4 earnings with $81.3 billion in revenue and 17% growth, but shares fell 12% after CFO Amy Hood cited AI hardware constraints limiting Azure's scaling potential.
- Meta posted stronger Q4 growth at 21% year-over-year with $59.9 billion in revenue and stock surged 10%, rewarding Zuckerberg's plan to ship AI models and integrate them into video creation and ads.
- Investors punished Microsoft for supply-side constraints despite OpenAI ownership while rewarding Meta's $30 billion-plus AI infrastructure spend before shipping consumer products, signaling the market values execution speed over current model pedigree.
Summary
Microsoft posted strong Q4 earnings with revenue of $81.3 billion, beating consensus of $80.2 billion, and 17% year-over-year growth. The stock dropped 12% after the earnings call anyway, down 17% over six months. CFO Amy Hood said directly that limited AI hardware availability is capping Azure's growth rate and revenue potential. Microsoft owns 27% of OpenAI's new for-profit entity and has competitive models through Codex and Anthropic deals, but none of that matters if the cloud business cannot scale fast enough to serve demand.
Building more data centers takes time. Beyond capacity constraints, questions loom about energy bottlenecks, chip availability, and whether TSMC is investing enough on the CapEx side to keep pace if AI continues accelerating. If TSMC stumbles, Microsoft and others may need to pressure Samsung and Intel to step up.
Meta had the opposite earnings reaction. Q4 revenue came in at $59.9 billion, beating expectations of $58.5 billion, with 21% year-over-year growth—faster than Microsoft's top line. The stock popped 10% after hours, pushing market cap to $1.84 trillion. Mark Zuckerberg framed 2025 as a year of rebuilding AI foundations through acquisitions, hires, and restructurings. He signaled the company will ship new models and products in coming months, with early versions showing a "rapid trajectory."
The real test is not whether Meta lands a frontier LLM but what it does with the models inside its business. Image and video models are likelier to drive near-term value than an LLM, since they slot naturally into content generation, ads, and targeting across Instagram and Facebook. An LLM can seep into product details more subtly. YouTube already uses AI summaries that let viewers extract parts lists from PC build videos or ask Gemini to summarize comment sentiment.
Meta owns Manus, a product team known for building agents. Zuckerberg can integrate AI into short-form video creation through prompts that let creators describe a video, generate B-roll, and remix content. That remix functionality is underrated. Character swaps and format remixing lower the friction for creators without true inspiration to produce engaging new content.
The market's read is clear. Meta is spending over 50% of revenue on AI infrastructure and has not shipped a single compelling consumer AI product yet. Investors rewarded the ambition and growth rate anyway. Microsoft, despite its OpenAI upside and model pedigree, got punished for being constrained on the supply side.