Logitech's MX Keypad: The AI Idea It Already Gave Up On

Logitech wants to be the hardware layer for AI. Last year it tried the cheap way: a free tool that turned your mouse and keyboard into a shortcut to ChatGPT. In June, it quietly killed that tool. Now it's selling you the shortcut instead.

The new device is the MX Keypad, the company's pitch to software developers. It sits next to your computer as a pad of nine little screens, each a button you can relabel and reprogram. Press one and it fires an app, a command, or a chunk of AI you've set up in advance. It auto-switches its layout to match whichever code editor you're using, and ships with native plugins for the tools developers actually live in: VS Code, IntelliJ, and GitHub Copilot. Logitech pitches the whole thing as an "AI control center".

Step back from the product name and something odd appears. Logitech's first AI product wasn't hardware at all. In April 2024 it released Logi AI Prompt Builder, free software that put a ChatGPT shortcut onto existing Logitech keyboards and mice. Two years later the feature was retired from the Options+ app. So a company that once gave the AI button away has now decided the same idea is worth a dedicated, premium piece of hardware aimed at the people who use AI the most and are the least likely to be impressed by a gimmick.

That reversal is the story. Not the keypad, but what it reveals about how a hardware company thinks it can ride the AI wave.

For a product company, the temptation is to conclude that AI is a feature you bolt on. Logitech tried that and, by retiring it, admitted the free feature didn't hold people. The new theory is the opposite: AI is abundant and cheap, so the scarce thing becomes the surface you use to control it — the physical layer of buttons and dials you touch to invoke and orchestrate the models across many apps. Charge for the surface, not the software.

It's a coherent idea, and Logitech has actually proved a version of it works. The MX Master 4 mouse, launched last fall, has already pulled in more than $100 million and become the fastest-adopting product in company history. The pitch there was the same: a physical, tactile control on top of software that would otherwise live in menus. A pad of programmable keys is that thesis taken further up the stack.

But here is where the investor should pause. The MX Keypad is not a new category. It is a competitor to the Elgato Stream Deck, the pad of keys that has dominated this space for years and is now being marketed by Elgato as exactly what Logitech is describing — an "AI controller", a physical device to manage agents and apps with real keys instead of window-switching. Logitech is arriving at a crowded party and calling the appetizers by a new name. That is positioning, not invention.

The economics matter for the same reason. The MX Keypad lands in Logitech's Keyboards & Combos bucket, the slowest-growing slice of the company's most recent quarter — up just 2% while pointing devices grew 16%, gaming 12%, and video collaboration 11%. A $200-ish keypad for developers is a real business if you're Elgato, a rounding error for a company that did $1.23 billion in a single quarter. The AI narrative and the AI revenue are different things, and the keypad quietly belongs to the narrative.

Here is the deepest problem, and it's the one nobody in a press release will tell you. These devices only create value if the buyer does the work. Every honest review of Logitech's earlier MX Creative Console — the same concept pointed at video editors instead of coders — said the same thing, captured in CNET's verdict that it's useful "if you put the work in". The pad is a blank slate until you configure ninety-odd keys and keep them maintained as your tools change. A shortcut you must build and upkeep yourself is not a shortcut for most people. It rewards a small, obsessive group.

Which is fine, if you know that's the business. A power tool for devoted users is a good business. The mistake would be to read the keypad as evidence that AI is turning into Logitech's growth engine. The category is small, dominated by an incumbent who already owns the name, sitting inside the company's slowest product line, and its value depends on a behavior — sustained, manual configuration — that most users will not sustain.

So the test to watch is not the launch. It's what the device does to a retreating buyer. Does a developer keep the MX Keypad and keep using its Copilot keys after the novelty of buying it wears off? That behavior — return rate, and whether owners rebuild their profiles as their workflow shifts — is the only number that tells you whether Logitech found a real thing or just a premium way to restate a free idea it has already given up on once.