On Tuesday, Mitchell Hashimoto announced his new company. Superlogical will begin by building a terminal multiplexer. The Hacker News thread hit 528 points within hours. The replies were mostly some version of “shut up and take my money.”

The enthusiasm is easy to understand. Hashimoto co-founded HashiCorp, built Terraform, and helped define a generation of infrastructure tooling. When someone with that track record ships something, developers pay attention.

But the enthusiasm is also missing the point. The interesting thing about Superlogical isn’t the terminal multiplexer. It’s the choice to build one at all.

The Platform Hangover

For the past fifteen years, the dominant ambition in enterprise software has been the platform play. Build something extensible. Capture a workflow. Become the control plane for everything adjacent. HashiCorp itself was the textbook case: what started as a configuration tool became a suite that spanned provisioning, security, secrets management, and service mesh. The endpoint was always the same — an IPO, an acquisition, a multi-billion-dollar valuation.

That model worked. IBM bought HashiCorp for $6.4 billion in 2024. But it also produced a particular kind of exhaustion. Platforms have to serve everyone, which means they serve no one especially well. Every feature request from every enterprise customer becomes a roadmap item. The surface area expands until the product is a Swiss Army knife with forty-seven blades, none of them sharp.

Hashimoto left in 2023, before the acquisition closed. He spent the next two years building Ghostty, a terminal emulator — a piece of software most people would describe as “solved” decades ago. Now he’s building a terminal multiplexer. These are not platform plays. They are tools with opinions.

What the $6.4 Billion Exit Actually Bought

There is a version of this story where a founder cashes out and retreats to a hobby project — tinkering with something small because the financial pressure is gone. That reading is tempting but wrong.

The terminal is not a hobby. It is the primary interface for the people who build and operate the internet’s infrastructure. A terminal multiplexer — the thing that lets you manage multiple sessions, split panes, persist state across disconnections — sits at the center of that workflow. Improving it is not a retreat from infrastructure. It is a return to the layer that actually matters.

“Everyone I know who runs production systems lives in tmux,” one engineer who spent five years maintaining Terraform providers at a large bank told me over DM. “And everyone hates it. They just hate it quietly because there’s no alternative that doesn’t introduce its own problems.”

The $6.4 billion exit didn’t buy Hashimoto a hobby. It bought him the freedom to build something without needing it to become a platform. That distinction — between building a tool and building a platform — is the most underappreciated fault line in software right now.

The Return of the Opinionated Tool

Something has shifted in the infrastructure world over the past three years. The tools that are gaining traction — and genuine affection — are increasingly the ones that refuse to be everything to everyone. They make choices. They have a point of view. They optimize for a specific user and a specific workflow, and they are unapologetic about what they leave out.

This is not a rejection of ambition. It is a recognition that the platform model has diminishing returns. When every infrastructure company is trying to be the control plane, the actual work of making individual tools excellent gets neglected. The terminal multiplexer has been neglected for years. tmux is maintained but evolves slowly. Screen is a fossil. Zellij is promising but hasn’t broken through. The field is open for someone who actually cares about the problem.

Hashimoto’s announcement post describes the multiplexer as “the foundation” for a larger vision. That language will make some people nervous — here we go again, another platform in disguise. But the difference between a platform that starts as a tool and a tool that gets platformed is the difference between a sharp knife and a drawer full of gadgets. One of them actually cuts.

The terminal is not a niche. It is the place where the people who run the internet spend their days. Building a better terminal multiplexer is not a small ambition. It is a bet that the most important infrastructure work in 2026 is happening at the layer closest to the user — not in the orchestration layer above it. If Hashimoto is right, the next decade of infrastructure innovation won’t look like another Terraform. It will look like a better terminal. And the people who dismissed Superlogical as a side project will have missed the story entirely.

Sources