OpenAI Smartphone
OpenAI Picks Foxconn's FIH Mobile to Build First AI Phone
Foxconn subsidiary FIH Mobile locked in the assembly order for OpenAI's first phone, packing a custom 2nm MediaTek Dimensity 9600 with dual AI processors. Three other Apple suppliers feed the component pipeline. Mass production targets late 2027 or 2028, a direct shot at breaking free from Google and Apple platforms.
Foxconn Wins the Assembly Gig
Foxconn subsidiary FIH Mobile has landed the assembly order for OpenAI's first smartphone, Chinese outlet Securities Times reported this week, citing supply chain sources. The Android Authority confirmed the report independently through its own supply chain checks. Neither OpenAI nor Foxconn has issued an official statement. FIH Mobile — a Foxconn unit listed on the Hong Kong Stock Exchange — designs, manufactures, and assembles phones for multiple brands. The two companies already partner on AI data center equipment, and Foxconn Chairman Young Liu said in late October 2026 that discussions about "additional joint projects" were underway, according to Gadgets360.
Under the Hood: The Spec Sheet
Analyst Ming‑Chi Kuo laid out the hardware blueprint months ago. Here is what builders should care about:
MediaTek appears to be the exclusive chip supplier, edging out Qualcomm for this socket, Android Authority reports. Kuo previously estimated 30 million units across 2027 and 2028 if development stays on track.
- Custom 2nm MediaTek Dimensity 9600 Built on TSMC's next‑gen node, per 2
- Two dedicated AI processing units Separate silicon just for on‑device inference
- LPDDR6 RAM The fastest mobile memory standard available
- UFS 5.0 storage Double the bandwidth of UFS 4.0 for model loading
- Camera hardware tuned for AI Sensors designed to help the model understand physical surroundings, not just take photos
The Apple Supply Chain Moves In
OpenAI is not building a supply chain from scratch. It is renting Apple's. Beyond FIH Mobile, three other Apple suppliers are feeding the component pipeline:
- Luxshare Precision — once tipped as the exclusive manufacturing partner by Ming‑Chi Kuo, now reportedly handling component supply instead of final assembly, per 1
- Lingyi iTech — structural and thermal components
- AAC Technologies — acoustic and haptic parts
The shift from Luxshare as exclusive partner to FIH Mobile as assembler suggests OpenAI is diversifying risk. EE Times noted earlier that Luxshare would serve as co‑design and manufacturing partner, but the latest reports from Securities Times and Cailian Press paint a wider supplier net.
Agents Replace the App Grid
The hardware is only half the story. The real bet is software. Instead of a grid of app icons, OpenAI's phone runs AI agents that complete tasks directly — no app‑switching required. Ming‑Chi Kuo shared a concept visualization showing an "agent task stream" replacing the traditional home screen, The Decoder reported. "Instead of apps, the smartphone could rely on AI agents to complete different tasks," Kuo wrote in his original analyst note, cited by.5 The phone runs a mix of small on‑device models for latency‑sensitive tasks and cloud models for heavier lifting. OpenAI's pitch: a phone that understands context continuously, anticipates what you need, and acts without making you tap through six apps.
The OS Wildcard
Nobody knows what operating system this thing runs. A customized Android fork is the obvious starting point — it brings an existing app ecosystem and developer tools. Building from scratch offers total control but zero apps on day one. 1 calls the OS "unknown" and notes the tradeoff clearly: "A customized version of Android is considered a natural starting point due to its app ecosystem. Building an OS from scratch would offer more freedom but face adoption challenges." For builders, the OS choice determines everything about the development surface. If it is Android‑based, the SDK story gets easier. If it is proprietary, OpenAI needs to ship a developer platform alongside the hardware. The EE Times report suggests OpenAI may tie hardware sales to subscription plans and build a developer ecosystem around AI agents.
Why OpenAI Needs Its Own Phone
This is not a hardware play for fun. OpenAI's entire business runs inside other people's platforms — Apple's iOS, Google's Android, and every browser. That means Apple and Google control permissions, background processing, sensor access, and default assistant slots. "By creating its own smartphone and hardware stack, OpenAI would be able to use AI in all kinds of features without restrictions," TechCrunch reported. Ming‑Chi Kuo put it bluntly: "The smartphone is the only device that captures the user's full real‑time state, which is the most important input for real‑time AI agent inference. Only by fully controlling both the operating system and hardware can OpenAI deliver a complete AI agent service," as quoted in.3 Sam Altman acknowledged the pivot on X, noting that while OpenAI once believed it could disrupt the iPhone with a new device category, it now sees the smartphone as central to AI use for years to come.
Timeline: Late 2027 or Bust
The clock is ticking. Cailian Press pegs market debut at late 2027 or sometime in 2028, according to Gadgets360. Earlier reports from Kuo suggested mass production could start in the first half of 2027, but supply chain timelines appear to have slipped. One wildcard: Apple sued OpenAI over alleged trade secret theft involving former employees who took confidential product development and supplier information. A federal judge granted Apple limited expedited discovery in October 2026, Android Authority reported. The lawsuit could slow hardware plans, but the fact that supply chain reports keep surfacing suggests the smartphone project is still moving. Google and Samsung are already shipping agentic AI features on their own devices. OpenAI needs to ship before the window closes.
Sources
- 1.Android Authority(androidauthority.com)
- 2.Gadgets360(gadgets360.com)
- 3.EE Times(eetimes.com)
- 4.The Decoder(the-decoder.com)
- 5.TechCrunch(techcrunch.com)
Nov 20, 2026
Arizona Opens Criminal Probe Into OpenAI Over Rogue AI Agents in Banks
Arizona Attorney General Kris Mayes announced a criminal investigation into OpenAI and other AI companies over the "apparent infiltration of banks by rogue AI agents." The probe comes as major banks including Morgan Stanley and JPMorgan Chase have already begun blocking AI agents from customer accounts.
Nov 20, 2026
Anthropic Shares Close Up 17% After Record $81 Billion IPO
Anthropic's stock closed its first trading day at $97.17, up 17% from its $83 IPO price, after opening at $111.50 and briefly touching a $2 trillion valuation. The $81.3 billion offering is the largest IPO in history, cementing the Claude maker as the first frontier AI lab to go public.
Nov 20, 2026
Perplexity Launches Orchestrator API for Model Routing and Failover
Perplexity opened its internal model-routing layer to developers with the Orchestrator API, giving one endpoint for OpenAI, Anthropic, Google, xAI and open-weight models with automatic same-model failover and unified billing. Unlike OpenRouter's 5.5% fee, Perplexity charges no platform fee on credits.
Related News
Nov 20, 2026
OpenAI Built a Codex-Powered System to Fix ChatGPT Slowdowns
OpenAI revealed an internal system called Performance Factory that uses Codex agents to investigate and fix ChatGPT performance regressions. Over three months it ran more than 1,700 investigations, and roughly 15 shipped fixes were pull requests written entirely by Codex. It is a working preview of how AI coding agents and human engineers can coordinate on production systems.
Nov 19, 2026
Perplexity Brain Creates 80,000 Agent Skills for 20,000 Users From Experience
Perplexity's Brain personalization system has written over 80,000 personal skills for more than 20,000 users, learning workflows from experience. The system reduced tool-call failures by 11% and dissatisfaction threads by 7.6% in production testing.
Nov 15, 2026
Perplexity Teammates AI Agents Arrive on Mac — Behind a $200 Paywall
Perplexity brought its Teammates AI agents to the Mac desktop app this week, but access requires the $200/month Max plan. The always-on agents, launched on the web a week earlier, can perform ongoing research, monitoring, and task execution independently.