Does AI make Money
retained text
Hey Guys, so I wanted to look at some hyperscalers to see how they make Money and start with MSFT Azure and MSFT 365 Cloud revenues are the only breakdown that matter to me because I am looking at AI revenues. For 2027 and 2028 revenue I took AI inputs (Ironically) which are wall st estimates so not sure how accurate these numebers are. I also dont care about how the revenues affects the valuation, **My only question is whether the AI revenue growth makes sense corresponding to AI CAPEX spend.** Note that MSFT only reports total CAPEX and the CAPEX for AI buildout might be around \~ 75% of total revenues: |**Fiscal Year**|**Azure Cloud Revenue**|**Microsoft 365 Cloud Revenue**|**Total Capital Expenditures**|**Year-End Total Cash Balance**|**Trajectory Notes**| |:-|:-|:-|:-|:-|:-| |**2028 (Projected)**|**$212.00 Billion**|**$128.00 Billion**|**$195.00 Billion**|**$58.00 Billion**|Azure approaches 2x the size of Office suite revenue| |**2027 (Projected)**|**$159.00 Billion**|**$113.00 Billion**|**$170.00 Billion**|**$52.00 Billion**|Model-as-a-Service (MaaS) token volumes double| |**2026 (Actual)**|**$115.10 Billion**|**$100.30 Billion**|**$115.95 Billion**|**$76.84 Billion**|Azure officially crosses over M365 Cloud line| |**2025 (Actual)**|**$75.30 Billion**|**$88.90 Billion**|**$64.55 Billion**|**$94.57 Billion**|Historical baseline period| |**2024 (Actual)**|**$56.20 Billion**|**$77.30 Billion**|**$44.40 Billion**|**$75.54 Billion**|Historical baseline period| **My conclusions:** 1. MSFT 365 Cloud revenues aren't really accelarating at all, does that mean Co-pilot is useless? The company that I work for uses Co-pilot and I think its super useless, I wouldn't pay any money for that shit. I think this part of the revenue only includes MSFT Copilot revenue which is assistant for MSFT 365, but MSFT also has Azure AI Co-pilot agents whose revenues are included in azure revnues i beleive. In any case suprised MSFT/Wall St has priced almost no AI related MSFT 365 growth, if AI really increases productivity, this would be the first place where I would expect revenue acceleration. 2. Azure Revenues: Obviously this is the insane AI growth engine, but its hard to say how the revenues are broken down into renting servers /tokens/AI service. I took this Wall St Estimate not sure how accurate the breakdown is. Azure Full Year Revenue & Workload Evolution |**Product / Workload Component**|**FY24 Baseline (Est.)**|**FY25 Baseline (Official/Est.)**|**FY26 Baseline (Official/Est.)**| |:-|:-|:-|:-| |**Traditional Infrastructure (IaaS/PaaS)**|$42.30 Billion|$51.05 Billion|$62.10 Billion| |**Enterprise Databases & Data Pipelines**|$10.20 Billion|$14.25 Billion|$18.63 Billion| |**OpenAI Infrastructure Capacity**|$3.10 Billion|$8.25 Billion|$17.59 Billion| |**Organic Enterprise AI APIs**|$0.40 Billion|$1.45 Billion|$5.18 Billion| |**Total Standalone Azure Revenue**|**$56.00 Billion**|**$75.00 Billion**|**$103.50 Billion**| Organic Agnetic API's like the models they built for Morgan stanley/Walmart is only \~5 Bil of this revenue based on this? And Open AI capacity revenue is basically Open AI paying money to MSFT to rent their servers space but we dont know how Open AI will make money? The Traditional server rent is also increasing at a good pace but nothing spectacular right? I think the Iaas/PaaS Revenue is only related to renting their servers general compute/AI and also for API tokens. I thought the narrative was that MSFT has figured out how to monetize AI, and I would think that they were talking about a major part of revenue coming from enterprice agentic AI but that's not the case at all based on these numbers. **So why did the market reward MSFT after earnings? What am I missing? Feel like i definately am missing some revenues/ didnt breakdown revenues correctly from 10-Q cause this doesnt make sense to me.** Or does that mean that Market is pricing in expectation that after 2027 MSFT will start monetizing AI, just pure sepeculation that: 1. Agentic AI revenue will ramp from 5 Bil from other companies + 18 B from open AI and this will keep doubling => Basically its a bet on Open AI. 2. The Azure revenue is supposed to keep going up because companies will keep renting server space in the future for AI/ API revenue keeps going up? But there is no accelaration in these revnues in 2026 to prove this => Does 10-15 B in Revenue increase justify the AI spend. So in short my question what data points in 2026 revenues to the fact that AI revenue is accelerating to grow into 100's B's in 2027/2028 to justify CAPEX spend? It's absolutely okay to speculate but there must be some revenue growth accelaration at some point. Also the CAPEX is not one time investment we know that data centers depreciate over time .... **PS: To be fair agentic AI increase from 1 B to 5B is also insane growth => Maybe this the real AI growth engine long term companies using MSFT for differnet agentic AI work... And since they used open AI to make this models, this is a good proof point that Open AI models can be commercialized.** **So I guess its a bet that Open AI works and MSFT proved agentic AI business on a small scale of \~ 5 Bil but we need to assume Open AI can make \~ 100's of B using same business and that will be the AI growth engine for server CAPEX spend?**
document fetched 2026-11-08 00:35 · canonical https://reddit.com/r/ValueInvesting/comments/1xpwq3y/does_ai_make_money