TL;DR
- Custom Chip Partnership: OpenAI is said to be co-developing a smartphone processor with MediaTek and Qualcomm, with Luxshare handling system design and manufacturing.
- Volume Target: The unconfirmed program targets millions of annual units by 2028, which would rival Apple’s iPhone shipments.
- App-Free Design: The device would replace traditional apps with an AI agent layer that handles user requests directly through OpenAI’s models.
- Platform Gap: No new smartphone platform has succeeded since Android and iOS, and OpenAI lacks the carrier and distribution infrastructure incumbents control.
OpenAI is said to be developing custom smartphone processors with MediaTek and Qualcomm, targeting millions of annual units by 2028 with an agent-driven device designed to eliminate traditional apps entirely. None of the named partners have confirmed the plans.
Supply-chain analyst Ming-Chi Kuo of TF International Securities identified the partnership in an April 27 note. OpenAI, MediaTek, Qualcomm, and Luxshare have not issued official statements, and no regulatory filings or patent applications have surfaced to corroborate the program independently.
Supply Chain and Manufacturing Partners
Luxshare Precision Industry, an established Apple supplier, is said to be handling system design and manufacturing for the device. Luxshare assembles AirPods, Apple Watch units, and iPhone components, giving it deep experience with the tolerances and volumes required for consumer electronics at scale.
That manufacturing background would serve a fundamentally different product category under the OpenAI smartphone effort. Apple split iPhone manufacturing between Foxconn and Luxshare starting in 2020 to reduce single-supplier risk, and OpenAI appears to be adopting a similar two-vendor model from day one, with Luxshare on the smartphone and Foxconn on the Jony Ive device and data-center hardware.
Parallel product lines position OpenAI to compete across multiple hardware categories simultaneously, but they also double the coordination burden for a company that has not yet shipped consumer electronics at any scale. Coordinating separate chip architectures, firmware stacks, and supply-chain contracts with different manufacturing partners requires dedicated hardware-program management teams that OpenAI would need to recruit or build from scratch.
Luxshare’s assignment to the smartphone contrasts with a January 2026 shift in which OpenAI moved production of its first AI-powered consumer device from Luxshare to Foxconn’s facilities in Vietnam. Luxshare’s reappearance as the manufacturing partner for a separate smartphone program confirms the parallel-track approach. Foxconn, which confirmed it would build AI servers for the Stargate project at its Ohio plant, remains tied to OpenAI’s data-center and first-device efforts while Luxshare takes the smartphone assignment.
OpenAI’s planned consumer device, designed with Jony Ive, was described as a non-phone form factor when the team internally unveiled the finalized prototype in late 2025. A smartphone targeting volumes an order of magnitude larger than any single-purpose AI gadget marks a sharp pivot from that original vision.
OpenAI’s broader five-device hardware roadmap, disclosed in January 2026, already included products stretching toward 2028, but no prior disclosure identified a smartphone-class device at this scale. OpenAI’s Sweetpea earbuds, scheduled for September 2026, would represent the first hardware product OpenAI delivers to consumers, well before the smartphone could reach production.
Scale Targets and App-Free Vision
At 300 to 400 million annual units, the planned shipment target would place OpenAI in direct competition with Apple’s iPhone volumes. Reaching that scale would require not just competitive hardware but a global distribution network, carrier partnerships, and manufacturing capacity that OpenAI has not operated at any comparable level.
Samsung, the world’s largest smartphone maker by unit volume, coordinates dozens of factories and carrier agreements across more than 100 markets to sustain its annual output. Building equivalent infrastructure from zero in under two years would be unmatched in the consumer electronics industry.
The reported user-facing concept is an app-free smartphone built around real-time AI agent inference. Rather than running a conventional app ecosystem, the phone would route user requests through OpenAI’s models, which would interact with services directly on the user’s behalf. That design would bypass the app-store economics that generate tens of billions in annual revenue for Apple and Google, potentially attracting developers frustrated by platform fees while simultaneously removing the distribution mechanism those developers rely on today.
Removing the app layer creates a structural barrier to adoption that no major platform has willingly chosen. Every smartphone platform that has gained meaningful market share, from iOS to Android to the now-defunct Windows Phone, relied on a third-party app ecosystem as its primary retention mechanism. OpenAI’s proposed model would need an alternative stickiness engine, presumably the quality and breadth of its AI agent layer, that persuades users to abandon app libraries accumulated on iOS since 2007 and Android since 2008.
Offline functionality, regulatory compliance for financial and health services, and backward compatibility with existing workflows remain open questions. Banking and healthcare apps are subject to jurisdiction-specific data-residency and encryption mandates that an agent-mediated architecture would need to satisfy before carriers and regulators could approve the device for sale.
Capital constraints compound the execution challenge. OpenAI is once more pursuing additional capital as the company’s operating costs outpace revenue by a wide margin. Building a smartphone supply chain from scratch, including component procurement, carrier certification, after-sales support, and warranty infrastructure, would require sustained capital investment over multiple years before any device reaches consumers.
OpenAI’s recent organizational changes, including a leadership reshuffle in April 2026 that saw Kevin Weil depart and Greg Brockman take interim oversight of products, add execution-risk questions for a multi-year hardware program of this ambition.
Competitive Context and Outlook
OpenAI would enter a smartphone market where AI integration is now standard equipment. Apple ships Apple Intelligence across its iPhone lineup, Google embeds Gemini into Pixel devices, and Samsung bundles Galaxy AI with its flagship phones. All three control their own silicon, operating systems, and app ecosystems, giving them distribution advantages that a newcomer cannot replicate in a two-year window.
Each also maintains direct relationships with hundreds of mobile carriers worldwide, a channel that takes years to establish and that no AI startup has built.
Standalone AI hardware has a troubled track record. Humane’s AI Pin launched in 2024 to poor reviews and the company was acquired by HP in 2025 after failing to gain traction. Rabbit’s R1 shipped but saw rapidly declining consumer interest once the novelty wore off. Both devices attempted to replace the smartphone rather than compete with it directly, and neither achieved meaningful commercial scale.
OpenAI’s approach differs in targeting the smartphone form factor itself, competing on the same physical ground as incumbents. No company has launched a viable new smartphone platform since Google acquired Android in 2005 and Apple introduced the iPhone in 2007, a two-decade gap that underscores the difficulty of the undertaking.
Ecosystem lock-in reinforces that barrier. By 2026 Samsung, Xiaomi, and Oppo rely on Google’s Play Services, Maps, and Gmail integration to ship over one billion Android devices annually.
By 2028, OpenAI would need either a licensing agreement with Google, which would limit its agent-driven ambitions, or a comparable alternative OS layer built from scratch.
MediaTek and Qualcomm bring relevant capabilities to the effort. MediaTek completed tape-out of its first 2nm flagship SoC in late 2025, while Qualcomm is preparing its own 2nm chipset for late 2026. Both companies have deep experience designing mobile processors optimized for on-device AI inference, a requirement for the continuous agent architecture the reported device envisions.
Qualcomm’s recent push into AI data-center inference chips signals a broader strategic alignment with large-model operators like OpenAI. For both chipmakers, co-developing a custom SoC for an entirely new platform owner would diversify their customer base beyond Samsung and Xiaomi at a time when their traditional Android OEM customers face slowing upgrade cycles.
OpenAI has not confirmed the smartphone program, and the 2028 mass-production timeline leaves roughly two years before any device could enter carrier-certification testing. The Sweetpea earbuds, scheduled for September 2026, will provide the first public data point on whether OpenAI can deliver consumer hardware on time. That result may determine whether MediaTek, Qualcomm, and Luxshare commit final production capacity to the smartphone program.