𝗖𝗹𝗮𝘂𝗱𝗲 𝗢𝘃𝗲𝗿𝘁𝗮𝗸𝗲𝘀 𝗖𝗵𝗮𝘁𝗚𝗣𝗧 𝗶𝗻 𝗨𝗦 𝗕𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗦𝗽𝗲𝗻𝗱
Watch the bill instead of the benchmark. Three major updates changed the AI landscape this week.
Claude leads US business AI spend
Ramp's AI Index tracks payments from 40,000 US businesses. Claude now holds 34.4% of AI spend. ChatGPT holds 32.3%. This is the first time Claude leads in spend.
Note the difference between money and seats. IDC data shows only 19% of teams use Claude deeply. Spend shows where money goes. Seat count shows who logs in. Do not confuse the two.
Claude Code billing changes
Claude Code agents move to a separate credit pool on June 15. This cost ranges from $20 to $200 per month.
The following use the new pool:
- Agent SDK
- claude -p
- Claude Code GitHub Actions
- Third-party agents like Zed and Conductor
These use full API rates. There is no rollover. Audit your scripts now. A loop running claude -p costs nothing today. It will cost money soon.
Microsoft ships new models
Microsoft released seven self-built models. They did not use OpenAI distillation.
Microsoft claims:
- Testers preferred MAI-Thinking-1 over Claude Sonnet in blind tests.
- The models scored 53% on SWE Bench Pro.
These are Microsoft's own numbers. Independent verification is missing. Wait for third-party benchmarks before you move your workloads.
How does the Claude Code billing change affect your workflow?
Optional learning community: https://t.me/GyaanSetuAi