OpenAI’s smart speaker bet looks like a money pit

OpenAI may be preparing its first in-house hardware product, but the rumored device is heading into a market that has already cooled.

Earlier this week, Bloomberg’s Mark Gurman reported that OpenAI’s debut first-party device would be a “humanlike” rechargeable smart speaker. The report lands just months after former OpenAI executive Fidji Simo warned employees this past spring that the company risked missing its moment because it was being “distracted by side quests,” including costly software projects such as Sora, which OpenAI shut down in April.

That warning looks especially relevant given the state of the category. Jitesh Ubrani, research director for worldwide device trackers at IDC, told Engadget the smart speaker market has been slowing for years. Shipment volume, he said, has fallen to “tens of millions units” after 2022, when tech giants led by Amazon shipped more than 100 million devices.

IDC’s figures show the downturn clearly:

• shipments fell 16.3 percent year over year in 2023
• they dropped another 11.8 percent in 2024
• the market still contracted 6.7 percent in 2025, even after Alexa+ launched
• IDC expects another 9.6 percent decline by the end of this year before the category flatlines next year

Ubrani said most buyers simply do not need to upgrade. Features are increasingly cloud-based, while the core use cases remain basic: music, podcasts, timers, and other simple tasks. That makes it hard for any newcomer to break through, especially one trying to sell a more complex and likely pricier device.

Recommended reading

Vietnam taps RusHydro to upgrade Soviet-era dams

According to Bloomberg, OpenAI’s speaker would include a camera, sensors, a rechargeable battery, and mechanical elements that can move on their own. Ubrani noted that the best-selling smart speakers have historically cost under $100, and consumers generally do not see the category as premium. A $300 speaker, he said, would be a tough sell without cheaper options.

OpenAI reportedly believes the device’s key differentiator will be its personality and its ability to interact with users on a humanlike level. It would rely on the company’s GPT-Live-1 voice model, built on a duplex architecture that can process inputs while generating outputs. OpenAI has said that setup enables more natural back-and-forth conversation, better timing, and live translations.

But Amazon made similar promises with Alexa+ in 2025, including mood detection, contextual awareness, and memory of earlier parts of a conversation. Ubrani said those upgrades have not materially changed the market’s trajectory.

Smart speaker economics remain weak

Even if OpenAI can ship the product, making money from it is another matter. Engadget points to Amazon’s long struggle to monetize Alexa. A Business Insider report said that as of 2022, Amazon was still selling many Echo devices at cost and had not found a durable business model for routine assistant interactions such as weather requests or playing music. In the first quarter of 2022, Amazon’s Worldwide Digital unit — including Alexa, Echo devices, and Prime Video — reportedly posted an operating loss of $3 billion, with the “vast majority” tied to the digital assistant.

Amazon has since said that 1 million people were using Alexa+ as of June 2025, but it has shared few other major milestones. As Ubrani put it, companies mainly use these devices as a way into their broader ecosystems.

There is still a strategic case for OpenAI getting into hardware. Ubrani told Engadget that if AI systems are meant to understand the world around users, they cannot remain confined to a browser. A smart speaker is also easier to launch than a smartphone or smart glasses, both of which come with more complex supply chains, and in phones' case, carrier certification.

Even so, the risks are obvious. Meta abandoned its Portal smart displays in 2022, the same year it laid off 11,000 people. OpenAI, meanwhile, remains deeply unprofitable. Leaked financial documents show it lost $5.09 billion in 2024 and $38.5 billion in 2025, and the company has said it does not expect profitability until at least 2030 because of rising AI costs.