One Way Street
We are still early in the adoption phase of AI. The first and most developed business use case is clearly for writing code. The tools are powerful and production ready, and most software developers have used AI in their coding, if not for work, at least for a weekend project. In a survey from The Pragmatic Engineer, 95% of respondents use AI tools at least weekly, and 75% use AI for at least half of their software engineering work. Stack Overflow’s 2025 survey found that 50% of professional developers use AI tools daily.
Last October, before the meteoric rise of Anthropic’s Claude Code, Menlo Ventures published data showing 3.2x YoY growth in enterprise AI spending, from $11.5B in 2024 to $37B in 2025.
The way to get real productivity gains from AI coding tools is to let the AI write code for you. And once you start letting the AI write code for you, you quickly wind up with a codebase that you don’t really know how to manage by hand. So adopting AI in software development is powerful but involves giving up control.
It certainly seems like a one-way street.
Today, The Information reports on how Anthropic is now leading the way in extracting a price for that dependency:
Businesses whose employees are heavy users of Anthropic’s Claude products are likely to pay significantly more for them, as the company changed its pricing model to charge enterprise customers based on the amount of AI they use rather than just charge flat fees.
Many business appear willing to pay the costs:
Many technology firms and other large Anthropic customers say they plan to eat the soaring costs as they try to boost productivity among software engineers and salespeople by automating certain tasks.
But as good as Claude is, it is not the only option on the market. Switching from Claude Code to Codex is not only possible, it’s easy. And there is a robust ecosystem of open source models that will start to become feasible replacements for the frontier models over the next few years for many coding tasks.
And as Anthropic, OpenAI, Google, and others begin to charge more, businesses will look to find ways to relieve the pressure on their margins.
My husband's company tracks AI usage. When he told me that, I - as someone who is tech-adjacent but in marketing - absolutely assumed they were tracking it in a bad way. We had just completed a hiring process where we screened out a lot of users whose cover letters came up as x percentage AI-generated, so from my perspective, a company would only be tracking this as a negative measure.
I was totally wrong - they were tracking it because their engineers who were not using it were not as productive! They wanted them to use more AI. I was, and remain, a bit floored. My job is still very insulated from AI. But my partner LOVES it - it's fun for him, and he gets more work done. He likens it to conducting an orchestra instead of playing an instrument, though I don't know how far that metaphor stretches.
Meanwhile, my teenagers are so anti-AI that they are practically boycotting my husband's in-home office. The algorithms that are feeding them are strongly against AI, to the point that one of them refused to watch my husband demonstrate a simple use case he had for scraping the PDFs of our church bulletins to determine which songs our choir had sung in the past three years. I thought it was fascinating. I'm not sure what use case I would have for it, but I haven't found one yet, so I guess I'm safe from being boycotted personally. Maybe. I'm sure they can find another pathway there :-)