MICROSOFT BAN AI IN ORGANIZATION

**Microsoft** has reportedly begun canceling most of its direct Claude Code licenses, and moving engineers toward using GitHub Copilot CLI. That comes just six months after the firm first opened up access to Claude Code, encouraging thousands of its developers, project managers, designers, and other employees to experiment with coding. The tech became popular fast. Perhaps too popular. The scale at which employees use it is now prompting the firm to reverse course on a tool its own engineers had come to rely on.

HERE’S WHY

The biggest hidden driver is simple platform politics. Microsoft has invested upwards of $13 billion into OpenAI and completely owns GitHub (the makers of Copilot).

When internal teams use external, unoptimized pay-as-you-go APIs, that cost registers as an immediate operational expense.

Essentially, they are paying twice: billions to build the house, and then millions out-of-pocket to eat at a restaurant next door. By forcing developers back onto their own infrastructure via GitHub Copilot, they convert a volatile, unpredictable cash expense into usage that justifies the massive capital investments they’ve already committed to the balance sheet.

And yes AI cost is high so does revenue ,Comparing Microsoft’s financials from the “Pre-AI Model” era to today reveals a fascinating paradox

If you look purely at Microsoft’s standard operating expenses, their margins look better than ever. However, if you look at their Capital Expenditures (CapEx) — the money spent building the massive data centers and buying the NVIDIA chips required to run AI — the efficiency ratio tells a completely different story.

The real financial strain isn’t appearing under standard “Expenses” on the income statement; it’s hidden in Capital Expenditures (CapEx) on the cash flow statement.

In 2022, Microsoft only had to spend roughly $1 for every $8.70 of revenue they brought in to keep their data centers updated. Today, they are spending $1 for every $2.68 of revenue. In a single recent quarter, Microsoft poured a staggering $30.88 billion purely into physical AI infrastructure.

This massive infrastructure buildout explains exactly why management pulled the plug on third-party tools like Claude Code

And In future token cost will go down as technology improved and capExp efficiency will go up coz when u invest in research u don’t get instant result or maybe i’m wrong in this case .so we don’t know for sure since still there are lot of experim.. and research is going on

while china deepseek reduce 75% cost and trained model without using nvidia chips https://substack.com/@mactek/p-199288506