Humane and Rabbit proved that an AI gadget without an ecosystem dies fast. OpenAI’s rumored phone matters because it looks like a bet that only a company with ChatGPT scale distribution can make AI hardware stick.
Two companies tried to build the future of AI hardware. Both failed.
The Humane AI Pin was permanently disabled on February 28, 2025, after HP acquired its remnants for $116 million and shut down the servers that ran it. Every device that had shipped became a brick. The Rabbit R1 retained only 5,000 active users five months after launch. Roughly 95 percent of the 100,000 people who pre-ordered had stopped using it. Those failures matter because they exposed a basic problem with AI-first gadgets: novelty is not an ecosystem.
OpenAI is now trying the opposite bet. According to Ming-Chi Kuo, the company is working with MediaTek and Qualcomm on smartphone processors, with Luxshare as its exclusive co-design and manufacturing partner, and is targeting mass production in 2028. Kuo told investors the device could reach 300 million to 400 million annual units if it works, a scale The Next Web reported would exceed iPhone volumes. The interesting part is not the rumor that OpenAI wants a phone. It is the strategic admission underneath it: the AI layer was not enough, so now the lab wants the device too.
That is the thread connecting OpenAI to the two AI hardware flops that came before it. Humane and Rabbit tried to wrap an assistant around a gadget and hope the gadget created its own demand. It did not. People already had a screen in their pocket, an app store, a payments system, and years of learned behavior sitting inside the phone they owned. A new AI device had to beat all of that at once.
OpenAI at least starts with a real distribution advantage. TechCrunch reported that ChatGPT is nearing one billion weekly users, giving the company something Humane and Rabbit never had: a large installed habit. That still leaves the hard part. OpenAI has no retail footprint, no carrier relationships, and no hardware brand that consumers already trust. A huge user base can get a device launched. It does not guarantee people will keep reaching for it.
Qualcomm's role shows why this story is bigger than one possible handset. Bloomberg reported that Qualcomm shares surged on Kuo's report, and The Next Web reported the jump reached as high as 13 percent in premarket trading before the company's April 29 earnings report. Investors are treating OpenAI as a possible escape hatch for a chip supplier that already sits inside the mobile stack. If OpenAI ever controls the interface where users talk to software, Qualcomm is not just selling components into one more phone. It is helping build a new platform that could pressure the companies that currently define mobile computing.
Luxshare sits in an awkward spot too. MacRumors reported that Luxshare is the manufacturing partner named in Kuo's supply chain account. Luxshare already assembles iPhones for Apple, which makes the pairing notable without pretending Apple depends on it for most production. If Kuo is right, one of Apple's manufacturing partners is also helping a would-be rival prototype the next version of AI hardware.
The public signal from inside OpenAI was subtle but hard to ignore. On April 26, Sam Altman wrote on X that it was "a good time to seriously rethink how operating systems and user interfaces are designed". That reads differently when it lands one day before a supply-chain analyst says OpenAI is building a phone. It also fits the company's earlier move into custom infrastructure: last October, OpenAI and Broadcom said they would deploy 10 gigawatts of OpenAI-designed AI accelerators. The company is no longer acting like software will be enough.
The obvious caveat is time. Kuo's timeline points to 2028 mass production, which is far enough away for the plan to change, slip, or die. Humane and Rabbit still matter because they show how brutal this category is once the demo gives way to daily use. OpenAI may have the first AI service with enough reach to make a serious device launch plausible. It still has to prove that vertical integration solves more than the last two companies' marketing problem.