The CTO Circle Newsletter · June 23, 2026
This is Marco from CTO Circle. My goal is to deliver the most value in the fewest words.
Let me know how to improve. I respond to every reply.
In this issue
What happens when AI stops being subsidized, the AI talent wars, infrastructure designed for agents, community notes, and open engineering leadership roles.
Sponsored - SPONSOR CTO CIRCLE HERE
Fix what's breaking in your AI agent
Latitude automatically detects all the ways your agents fail in production, and gives your coding agent the tools to fix it.
Use code CTO CIRCLE for 50% off your first 3 months.
Short opinionated trend
The AI bill is coming
One of the best conversations from our last CTO Circle gathering in Madrid was about what happens when AI stops being subsidized.
For the last few years, a lot of AI adoption has been hidden behind flat subscriptions, VC-funded infrastructure and very generous pricing. That phase is ending. The moment AI stops looking like €20/user/month and starts looking like actual infrastructure usage, finance teams will wake up very fast.
Most AI workloads do not need frontier models. That is why inference and AI cloud companies are so hot right now: Baseten, Fireworks, Together, Groq, Cerebras, Replicate, Modal, Nebius, DeepInfra, CoreWeave. They are all attacking different parts of the same problem: how to run AI workloads faster, cheaper and with more control.
When AI stops being subsidized, teams will need to be much more intentional about which models they use, where they run them, and how much each workload should cost. A smaller fine-tuned model can sometimes work better than a frontier model.
Classification, extraction, summarization, routing, internal search, first-pass review, and support triage are valuable workloads, but they should not all have the same cost profile.
News, reads & launches
Trump says he no longer sees Anthropic as a national security threat, as the politics around frontier AI labs continue shifting from model risk to industrial strategy.
Anthropic is hiring Nobel laureate John Jumper, while OpenAI is bringing in Noam Shazeer and Dean Ball. The race is now as much about talent, policy, and market structure as model releases.
Cloudflare launched temporary accounts for AI agents, letting agents deploy Workers without signing up first. Agentic workflows need infrastructure designed for machines, not human dashboards.
Accenture is buying Dragos, runZero, and NetRise in a $4.2B cybersecurity deal covering OT security, attack-surface visibility, and firmware supply-chain data.
There is minimal downside to switching to open models. As proprietary models become more constrained, open-weight models start looking less like a compromise and more like a practical fallback.
The Minimum Viable Unit of Saleable Software. If AI makes rebuilding software cheaper, which products are still valuable enough to buy instead of recreating internally?
The Agent Loop Architecture. Durable agents need loops, state, retries, observability, and infrastructure that can survive production workflows.
From the community
Yesterday we hosted our last CTO Circle gathering in Barcelona before taking a short summer break. We ended with an open discussion where everyone could bring the topics that were top of mind; this time, the conversation turned to junior engineering hiring and mentorship.
A community member also shared Claude Code Artifacts.
New here? Join the Slack community and introduce yourself.
Open engineering leadership roles
That's it for today. Let me know whether this issue gave you enough value, reply and tell me what to cut or add.
Best,
Marco