The artificial intelligence industry spent months anticipating OpenAI’s first hardware product.
However, it was reported Tuesday (July 14) that the screenless consumer companion developed with former Apple designer and io Co-Founder Jony Ive is still under development and will not ship before 2027.
The next day, OpenAI released its actual first hardware product. It’s not a consumer device. It’s a $230 programmable macro pad, called the Codex Micro, for developers managing AI coding agents.
The launch came days after Apple sued OpenAI, alleging the company used stolen trade secrets to accelerate its hardware push. OpenAI has denied the claims.
The Codex Micro was co-designed with boutique keyboard maker Work Louder. It sits beside a developer’s regular keyboard as a physical controller for Codex, OpenAI’s coding agent platform. It has 13 mechanical switches, a joystick and a rotary dial. Six illuminated Agent Keys display the live status of each Codex task. White means idle, green means unread chat, blue means thinking, pink means there is a question or user approval needed and red means error.
In a YouTube video, Work Louder Co-Founder Mike Di Genova described the Agent Keys as a “live view of your Codex threads.”
The joystick launches common workflows. The dial adjusts how much computing power an agent applies to a task. The product page called it a “command center for agentic work.” It comes in clicky and silent versions and is available while supplies last.
The Codex Micro is a niche product for power users, Axios reported Wednesday (July 15). Developers can track every agent without switching windows.
Keyboard Arrives as Codex Hits Its Fastest Growth Week Yet
On July 9, OpenAI merged Codex into the ChatGPT desktop app and launched ChatGPT Work as a new professional agent layer, according to a July 9 post on the OpenAI website.
More than 5 million people use Codex every week, OpenAI said the post. More than 1 million use it for work outside software development.
Codex and ChatGPT Work reached 8 million active users as of Tuesday, according to a post on social platform X by Tibo Sottiaux, Codex engineering lead.
Hello. We have reached 8M active users across Codex and ChatGPT Work.
We are once again resetting the usage limits for all. And we continue to not have the 5h rate limit as well, allowing everyone to explore the boundaries of GPT-5.6 Sol and discover how ambitious you can be.… https://t.co/pKunziOGRm
— Tibo (@thsottiaux) July 14, 2026
In a separate Tuesday post on X, OpenAI CEO Sam Altman called demand for GPT-5.6 Sol, the company’s newest model, “insane” and warned of infrastructure hiccups as OpenAI scales to meet it.
5.6 sol growth is insane.
the inference team has done heroic work to be able to support demand.
we are going to move mountains to continue to scale, but it is possible there are some hiccups soon.
— Sam Altman (@sama) July 14, 2026
First Commercially Viable AI Hardware May Not Be a Consumer Device
The Codex Micro is entering a category where failure is normal.
Humane raised $230 million and shut down after HP acquired the company in 2025. Rabbit R1 sold 100,000 units and faced widespread criticism when the product failed to match its demos. Both tried to redefine how consumers interact with AI. Both failed.
The Codex Micro takes a different approach. It does not replace an existing device. It extends an existing workflow. A developer already managing multiple Codex agents gets a physical surface that makes that work faster and more visible. The problem it solves did not exist two years ago.
Enterprise customers make up about 40% of OpenAI’s business, a share Chief Financial Officer Sarah Friar said she expects to grow to 50% by the end of the year.
The Jony Ive device is designed to answer a question that consumers are still forming. The Codex Micro answers a question developers are asking right now.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
The post OpenAI’s First Hardware Is a Keyboard, Not a Companion appeared first on PYMNTS.com.
Source link
