Add blog post: the AI coding subsidy multiple

content/blog/2026-06-23-open-control-plane-for-inference/index.mdx

@@ -6,7 +6,7 @@ authors:
   - name: "Bassam Tabbara"
     title: "Founder & CEO, Upbound"
     url: "https://github.com/bassam"
-    avatar: "/blog/open-control-plane-for-inference/bassam.png"
+    avatar: "/blog/authors/bassam.png"
     github: "https://github.com/bassam"
     linkedin: "https://www.linkedin.com/in/bassamtabbara/"
     x: "https://x.com/bassamtabbara"

content/blog/2026-07-03-ai-coding-subsidy-multiple/index.mdx

@@ -0,0 +1,89 @@
+---
+title: "Anthropic is subsidizing our AI coding at 20x. How long will it last?"
+description: "We measured what our team's Claude Code usage would cost at API prices. It runs about 20x our seat price on average, and 113x for our heaviest engineer. Here are the numbers, how we measure them, and the script to measure your own."
+date: "2026-07-03"
+authors:
+  - name: "Bassam Tabbara"
+    title: "Founder & CEO, Upbound"
+    url: "https://github.com/bassam"
+    avatar: "/blog/authors/bassam.png"
+    github: "https://github.com/bassam"
+    linkedin: "https://www.linkedin.com/in/bassamtabbara/"
+    x: "https://x.com/bassamtabbara"
+    bio: "Bassam is the founder and CEO of Upbound and the creator of Crossplane and Rook, two CNCF Graduated projects. He's spent the last two decades working on cloud and infrastructure, and is now bringing that work to AI inference with Modelplane."
+tags: ["inference", "cost", "open-source"]
+cover: "/blog/ai-coding-subsidy-multiple/cover.png"
+draft: false
+pinned: false
+---
+
+We are on Anthropic's Team plan at Upbound, and we are about as AI-pilled as a company
+gets. All of our engineers write code with Claude Code every day, on a bundled seat.
+It's a flat monthly price with limits, and you don't really see the token meter.
+
+Then one of our engineers switched to opencode and started running Opus 4.8 straight
+through the API instead of on his bundled plan. I looked at his usage a few weeks in.
+He was averaging about $5,500 a month. Same person, same work, roughly what he had been
+doing on the bundled plan at $125/mo. The only thing that changed was that we could
+suddenly see the token meter.
+
+That made me want to know what it would cost us if everyone moved off bundled seats and
+onto per-token pricing. It is not just hypothetical. Anthropic
+[moved its Enterprise customers to usage-based billing](https://www.theregister.com/2026/04/16/anthropic_ejects_bundled_tokens_enterprise/)
+earlier this year, tokens on top of the seat, and
+[shut off bundled-seat usage from other coding agents](https://thenewstack.io/anthropic-claudecode-opencode-split/).
+As they start optimizing for margins, it's clear to us that more policy changes are
+coming soon.
+
+## It's not easy to see the tokens on a bundled plan
+
+I was surprised by how hard it is to see the tokens consumed on a bundled plan. The
+analytics dashboards do not report tokens. The usage API that would report them needs an
+Enterprise account, as far as I can tell, which we do not have.
+
+Luckily, Claude Code writes a local log of every session, with token counts per message,
+and keeps roughly the last thirty days before it prunes. We wrote a small script that
+reads those logs and totals the tokens by model, priced at Anthropic's published rates.
+It emits only counts and costs, no prompts and no code, so people can run it and share
+the output without leaking anything. The script is
+[here](https://gist.github.com/bassam/27fa951ac01e0998b73b36d90297678b), and it runs on
+your own machine in about a minute.
+
+So we pulled the June numbers for 20 of our engineers.
+
+## The subsidy multiple
+
+Priced at list, the engineers' Claude Code usage ranged from under $20 to $14,177. About 90%
+of it is Opus, mostly Opus 4.8. These are floors, because the local logs had already
+pruned part of the month for most people by the time we looked. A few people were also
+out on PTO for a few weeks in June, so their months are partial, and we counted everyone
+anyway, which pulls the average down, not up.
+
+![Each bar is one engineer's June Claude Code usage priced at Anthropic list rates, shown with the multiple over a $125 Premium seat.](/blog/ai-coding-subsidy-multiple/subsidy-multiple.svg)
+
+Set that against Anthropic's Premium seat at $125 a month, and that's an average of 20x
+more expensive if we switched to paying for tokens. The average covers a wide range: the
+median engineer was about 11x, some barely touched it, and our heaviest ran $14,177 in
+June, 113x a single seat. I have started
+calling this the **subsidy multiple**: the tokens you burn in a bundled plan, priced at list, divided by
+what you pay for the seat.
+
+A few caveats: all prices are at list, not at a rate a big customer negotiates. A large
+share of it is cache reads from long agentic sessions, which bill at a tenth of the input
+price, so someone will argue the cost is softer than it looks. A bundled plan is also not
+all-you-can-eat: past the included limit you either get throttled or, with extra usage
+turned on, pay the overage at API rates. We have it on and barely touch it, a
+few hundred dollars a month, even as hard as we lean on Claude Code. Even after all of that, the shape does not move.
+
+## Owning our intelligence
+
+So we are doing the thing that follows from taking that seriously. Open-weight models are
+closing on the frontier faster than anyone expected. We've started looking at running our
+own models, where the cost is compute we control rather than a token price someone else
+sets and can reset whenever the market lets them. It also keeps our prompts and context
+ours. We're doing this on
+[Modelplane](https://github.com/modelplaneai/modelplane), our open control plane for
+inference, of course.
+
+It feels like a big correction is coming in AI, and we're planning for it. I'm curious
+what your subsidy multiple is.