When Anthropic Walked Into the Room With a Price Tag
Let me start with what actually happened in early 2026, stripped of the marketing veneer. Anthropic launched Claude for Work at $25 per user per month. Microsoft had already locked in Microsoft 365 Copilot at $30. The gap is five dollars. That five-dollar gap has done something remarkable: it forced the enterprise software industry to confront a question that exactly nobody in a position of power wanted to ask.
The question is simple. Are you actually getting what you paid for?
I spent eight years at McKinsey watching clients spend tens of millions on enterprise platforms. The pattern was always the same. Someone in technology had a thesis. A vendor had a solution. Budgets were committed. Pilots happened. Then adoption happened in that peculiar enterprise way where 20% of users drove 95% of the value while the other 80% periodically clicked around to justify the cost. The true return on investment lived in a conference room somewhere, discussed in a TECHanaly report commissioned by the vendor itself, never to appear again in an actual balance sheet.
The Math That Doesn’t Actually Add Up
Here is what makes this moment genuinely different. In January 2026, Microsoft commissioned a Forrester Total Economic Impact study on Copilot that claimed a 370% return on investment. That number got distributed across analyst briefings and sales decks. It sounded authoritative. It sounded independently validated. Then Gartner did something rarer: they actually looked at what enterprises were doing with these licenses in production.
Gartner found that 13% of enterprise Copilot seats were actively used on a daily basis.
Stop and sit with that number. Thirteen percent. I have sat through enough earnings calls and analyst presentations to know that when vendors talk about “adoption,” they mean something closer to “people who have been assigned a license and could theoretically use it if they felt motivated.” Daily active usage is the metric that matters. It tells you whether someone actually needed the tool or whether your organization paid for it anyway.
The math is unforgiving. If 13% of your Copilot seats generate the majority of that 370% claimed return, then the remaining 87% of seats are essentially paying for overhead. At $30 per seat per month, that overhead starts to feel expensive when someone else is offering comparable functionality for $25.
The Renewal Conversation That Changes Everything
Microsoft reported hitting 1 million paying enterprise seats for Copilot in their Q2 FY2026 earnings. Real milestone. They made a point of announcing it. What they did not discuss was the renewal rate from their initial cohorts, and that silence is precisely where the real story lives.
IDC surveyed 500 IT decision-makers in February 2026. Forty-four percent said they planned to run a formal competitive evaluation between Copilot and Claude for Work before their next renewal. Forty-four percent. That is not a rounding error. That is a structural crack.
Why would a company already committed to Microsoft 365 and all its integrations suddenly want to run a competitive process? Because the ROI conversation has become impossible to avoid. Your CFO is looking at her spreadsheet. She sees $30 per seat per month times your user count. She sees the Gartner number about 13% daily active usage. She starts asking her technology team why they are spending money on tools that 87% of assigned users barely touch.
And her technology team cannot answer that question with a Forrester report anymore. That approach stopped working the moment a credible competitor walked into the room with a lower price and comparable features.
The Unglamorous Part: Actually Competing on Value
This is the part that fascinates me most, because it is not sexy. Google Workspace with Gemini Business priced themselves at $22 per user per month and entered this market in Q1 2026. Morgan Stanley analysts called it the first genuine price war in the AI productivity suite category. The word “war” is doing work there, but what they mean is this: vendors can no longer escape the question of whether their tool is actually worth the money.
For years, enterprise software sold on switching costs, integration depth, and the assumption that bigger vendors were inherently safer. Anthropic is offering Claude for Work with admin controls and audit logs and the features that enterprise customers actually care about. They are not selling on ecosystem lock-in. They are selling on a different proposition: a capable tool at a lower price point, with less friction in evaluating whether it actually works for you.
That is a genuinely difficult market position to beat. Not because Claude is necessarily better at every task than Copilot. Not because Google is crushing both on features. But because the ROI conversation has become mandatory rather than optional. When that happens, the vendor who can deliver the feature set at the lowest cost while maintaining enterprise credibility tends to win the first wave of migrations.
What Actually Matters Here
The real insight underneath this five-dollar gap is that the AI productivity market is maturing. It is moving out of the early adoption phase, where customers trusted vendor claims because the technology was new and shiny, and into the phase where enterprises care about actual returns. They care about daily active users. They care about whether their employees are genuinely more productive or just newer at checking email.
This creates an opening for vendors willing to be honest about what their tool does and to price accordingly. Anthropic understood that most enterprises do not need enterprise-grade AI productivity tools at the Microsoft or Google scale. A solid, well-built, fairly-priced alternative would appeal to the 44% of IT decision-makers already having the uncomfortable ROI conversation with their leadership.
The five-dollar gap will likely close or reverse before the end of next year. Pricing in competitive markets compresses. What will not change is that Anthropic forced this conversation into the open. That forced conversation is the actual product they delivered to the market. Everything else is execution on a strategy that said: we compete on honesty about value, not on marketing claims about returns.
If you are in technology leadership and you have not yet run the numbers on your Copilot deployment, now is the moment. What is your actual daily active usage? What would a 10% or 15% reduction in per-seat cost mean for your budget? What would a competitive evaluation process look like in your organization? I would genuinely like to hear what you find.