The Curve is Bending
grantslatton.com · 3,027 words · saved by 1 readers
Predictions on near-term AI inference spending
The curve is bending We've finally crossed an inflection point of model utility where the output is actually worth more than the cost for real dev work. It's starting to make sense to throw lots of money towards inference. Some devs will likely get 5-digit AI inference budgets in 2026. There will probably be some disruption of the junior dev job market, or at least disruption of what it means to be a junior dev. This prediction is based on the last few months of using the newest generation of models and tools and feeling the productivity gains myself. For context, my AI inference spend for pro
saved by
related reading
- claude code is not making your product betterethanding.substack.com
- Don’t Outsource Your Thinkingteltam.github.io
- Shipping at Inference-Speed | Peter Steinbergersteipete.me
- How much I'm paying for AI productivity software (and the future of AI use)jacquesthibodeau.com
- My picture of the present in AI — LessWronglesswrong.com
- Malleable software in the age of LLMsgeoffreylitt.com
- Society's Technical Debt and Software's Gutenberg Momentskventures.substack.com
- Training Imperativesdan.io
- New Junior Developers Can’t Actually Code | N’s Blognmn.gl
- Where Are the Good AI Products?varunshenoy.substack.com
- Claude Code is the Inflection Pointnewsletter.semianalysis.com
- Coding Models Are Doing Too Much | whnrehiew.github.io