Happy Tuesday, fellow humans navigating the AI landscape! This week brought us some fascinating developments that remind us of an inconvenient truth: artificial intelligence, despite feeling like magic, is very much bound by the laws of economics, energy, and occasionally, common sense.
Let’s dig into what actually matters for your business.
Microsoft’s AI Revenue Reality Check
Microsoft reported that their AI services are now generating over $13 billion in annual revenue—a 175% increase year-over-year. Sounds impressive, right? It is. But here’s the part that should make business owners pay attention: the growth is coming primarily from enterprise customers who are integrating AI into existing workflows, not from flashy consumer products.
The takeaway? Companies seeing real ROI from AI aren’t the ones buying every shiny new tool. They’re the ones who identified specific, repetitive, time-consuming processes and applied AI surgically. One logistics company Microsoft highlighted reduced their invoice processing time by 70%—not by replacing humans, but by having AI handle the data extraction so humans could focus on exceptions and relationships.
If you’re still wondering “where do I even start with AI?”—start with your most annoying recurring task. The one that makes your team groan every Monday. That’s your pilot project.
The Great AI Power Struggle (Literally)
In news that surprises absolutely no one who’s watched their laptop fan spin into overdrive while running local AI models, the energy demands of AI infrastructure are becoming a genuine constraint. Data centers supporting AI workloads are now consuming electricity at rates that are making utility companies nervous and environmentalists furious.
Google, Amazon, and Microsoft are all scrambling for nuclear power agreements. Yes, nuclear. We’ve gone from “AI will solve climate change” to “AI needs its own power plants” in approximately eighteen months.
For small and medium businesses, this matters more than you might think. Cloud AI costs are likely to increase as providers grapple with energy expenses. If you’re budgeting for AI implementation in 2025, add a 15-20% buffer for potential cost increases. Also worth considering: some of the most practical AI applications—like the document processing mentioned above—can run on much smaller, more efficient models than the headline-grabbing large language models. Bigger isn’t always better, and sometimes “good enough” AI is actually the smart business choice.
OpenAI’s Enterprise Push Gets Serious
OpenAI announced expanded enterprise features this week, including better data governance controls, team management capabilities, and—finally—more transparent usage analytics. This is OpenAI acknowledging what Dr. White and I have been saying in consultations for months: businesses don’t just need powerful AI, they need manageable AI.
The new admin controls let organizations set usage limits by team, track which departments are actually getting value from their AI subscriptions, and implement approval workflows for sensitive use cases. It’s not sexy, but it’s exactly what IT directors have been begging for.
Here’s my consulting context: if you’ve been holding off on enterprise AI adoption because you couldn’t answer “how do we control this thing?”—that objection is getting weaker by the month. The tools are maturing. The question is shifting from “can we govern AI?” to “have we defined what governance we actually need?”
The Obligatory Funny Observation
I spent twenty minutes this week watching a colleague argue with an AI assistant about whether a meeting should be scheduled for “next Tuesday” or “Tuesday next week.” The AI was technically correct. The colleague was contextually correct. Neither would yield. The meeting was eventually scheduled via email, like it’s 2019.
We have built systems that can write poetry, analyze contracts, and generate photorealistic images, but calendar coordination remains an unsolved problem. I find this deeply comforting, actually. Job security for humans who understand that “let’s circle back” means “I’m ignoring this forever.”
Your Action Items
Before next Tuesday:
- Audit one recurring process in your business for AI potential (remember: annoying + repetitive = opportunity)
- Review your current AI tool subscriptions—are you paying for capabilities you’re not using?
- Have a conversation with your team about AI governance needs before you need them
Until next week, keep asking good questions and remain skeptical of anyone who says AI implementation is “simple.”
— Your friendly AI correspondent at On Your Side Technologies
