Microsoft announced it this week: 15 million paid Copilot seats. $5.4 billion in annualized AI revenue from the software layer alone. Azure AI contributing 13% of total cloud growth.
Those are not small numbers. They represent a decision by enterprises across every sector — healthcare, finance, logistics, manufacturing, professional services — to buy into AI at scale.
Now here's the number nobody announced: how many of those 15 million seats are actually doing anything useful.
The Seat Is Not the Solution
Buying a Copilot license is like buying a piece of heavy equipment. The machine doesn't build anything. The operator does. And the operator needs to know not just how to run the machine, but what to build, in what sequence, and how to spot when something's about to go wrong.
Most organizations that bought Copilot seats got the machine. They haven't found the operator.
What they need is someone who understands Microsoft 365 deeply — not as a user, but as a system — and who can design workflows that Copilot can actually execute reliably. That's different from knowing how to write a good prompt. Prompting is the easy part. The hard part is knowing which workflows are ready to be automated, which aren't, what the failure modes look like, and how to build the oversight layer that catches the misses before they become problems.
That skill set doesn't scale with the seat count. It doesn't come in the license.
Why This Is a Talent Problem
I've watched the same pattern play out with every major enterprise technology wave. ERP. Cloud migration. RPA. The technology arrives faster than the people who can implement it correctly.
This time is no different, except the stakes are higher and the timeline is compressed. Snowflake and OpenAI announced a $200 million partnership this week specifically to embed autonomous AI agents into enterprise data workflows. The architecture is moving fast. The talent pool that can work inside that architecture is not.
The people who can do this work — who understand Microsoft's Power Platform, who can build governance structures for AI outputs, who can sit between the AI capability and the business outcome and make them connect — are in real demand right now. They're not on job boards waiting. They're already employed, getting calls, and evaluating offers from companies that have made their AI priorities clear.
The Three Kinds of Companies Right Now
Companies that bought seats and did nothing. They'll spend 12 months explaining to their boards why the AI investment isn't showing up in the numbers. Eventually they'll bring in a consultant to tell them what they should have figured out at the start.
Companies that bought seats and deployed too fast. They turned Copilot loose on workflows it wasn't ready for, got inconsistent outputs, lost confidence in the system, and are now treating AI as a failed experiment. They didn't fail at AI. They failed at implementation sequencing.
Companies that bought seats and found someone who knows what they're doing. These are the ones that will be two to three years ahead of their competitors inside 18 months. Not because they had better AI. Because they had better humans running it.
What to Actually Do
If you're holding Copilot seats — or about to buy them — one question matters more than any product feature: who in your organization is accountable for making this work?
Not the IT department generically. A specific person or team, with actual expertise in Microsoft AI tooling, who owns the implementation, the governance, and the outcomes.
If you don't have that person, you're not running an AI program. You're running a subscription you're not using.
The talent exists. The Oracle layoff event two weeks ago sent 20,000 to 30,000 enterprise IT specialists to market. Some of them know Microsoft environments well enough to do this work. The window on that supply event is closing — 60 to 90 days before the best candidates are placed.
This isn't about being an early adopter. It's about not being the company that paid for the tool and got none of the return.
VC5 Consulting works with companies that need enterprise AI implementation talent — not theorists, but people who've done it inside real systems. If you're figuring out what your Copilot program actually needs to work, let's talk.