Here’s what’s on our plate today:

• 🧪 What OpenAI's phone rumors reveal about the future of the smartphone.

• 📰 Surveillance state's expo moment, Medicare's AI-ready overhaul, CEOs admit fuzzy gains.

• 🛠️ Weekend To-Do: test agent workflows, audit your apps, try on-device AI.

• 🗳️ Poll: What happens if OpenAI actually ships a phone?

Let’s dive in. No floaties needed…

The Laboratory

• Supply chain, not vaporware: OpenAI is working with MediaTek, Qualcomm, and iPhone assembler Luxshare on a custom AI phone targeting mass production by early 2027, built for continuous on-device inference.

• Intent replaces the app grid: The device uses AI agents to execute tasks directly, skipping apps entirely, which threatens Apple’s $109B App Store revenue engine and Google’s similar model.

• Incumbents aren’t standing still: Apple is embedding a 3B-parameter on-device model into its platform, and Google is replacing Assistant with agentic Gemini across 1B+ Android devices.

• The real stakes are control: If agents mediate every transaction, whoever operates the agent becomes the new gatekeeper, and an invisible middleman between you and your intentions raises questions that no one has yet answered.

What OpenAI’s phone rumors reveal about the future of the smartphone

Since the launch of the iPhone in 2007 and the introduction of app stores a year later, personal computing has been organized around a simple grammar: icons, taps, and discrete software containers that compete for attention. The model worked so well that it is estimated that 7-7.6B smartphones are in active use as of 2026. More importantly, it has defined how intent becomes action.

However, advances in technology are now challenging that grammar from multiple directions at once, most notably through the rise of artificial intelligence and, more specifically, AI agents capable of reshaping how human intent is translated into action across the digital world.

According to reports from supply chain analyst Ming-Chi Kuo, cited by outlets like 9to5Mac, OpenAI is fast-tracking an AI-first smartphone, targeting production as early as 2027, while simultaneously preparing for a public listing that could value the company near $1T. At the same time, Apple is entering a leadership transition, and Google is rebuilding its assistant layer around Gemini. The timing of these moves from some of the biggest names in tech is hard to dismiss as a coincidence. Which, in turn, raises the question of whether the interface that has quietly mediated digital life for nearly two decades is beginning to dissolve.

The supply chain says it’s real

There have been speculations about a hardware device from OpenAI for some time; however, this time, it is not a concept sketch. Kuo, whose supply chain intelligence on Apple hardware has made him one of the most closely followed analysts in the industry, reported in late April that OpenAI is working with MediaTek and Qualcomm on a custom processor, with Luxshare Precision Industry. This company also assembles iPhones as the exclusive manufacturing partner.

A follow-up note on May 5 accelerated the timeline, citing mass production as early as the first half of 2027, ahead of the 2028 window Kuo originally projected. The device would run on a customized MediaTek Dimensity 9600, built on TSMC’s next-generation N2P node, with a dual-NPU architecture designed for continuous AI inference.

Among the information making its way into media channels, two details stand out.

The headline specification is not the processor or the screen but the image signal processor, which features an enhanced HDR pipeline for what Kuo describes as real-world visual sensing. The phone’s defining input, in other words, is not a finger on glass but the environment around the user.

Second, Kuo identified OpenAI’s planned IPO as a driver for the accelerated schedule. A hardware product, even one that has not shipped yet, extends the company’s addressable market story beyond software subscriptions. OpenAI closed a $122B funding round in March 2026 at an $852B valuation and is targeting a Q4 2026 listing. The phone does not need to succeed as a product to succeed as a signal.

So why invest in a device at all?

To understand why OpenAI would invest in a device without being affected by its success, one needs to examine the concept underlying the hardware.

An operating system built on intent

According to TechCrunch, the device would rely on AI agents to complete tasks instead of apps. Which means that rather than opening a ride-hailing app, a restaurant app, a messaging app, and a calendar to meet a friend for coffee, the user would state the intention and the agent would coordinate execution across services.

This is not a new idea, but it is attracting new money. Nothing CEO Carl Pei argued at SXSW that smartphone apps will disappear, and future devices need interfaces designed for agents, not human-operated menus repurposed for automation. Nothing closed a $200M Series C to build exactly this.

On April 26, Sam Altman posted on X that “it feels like a good time to seriously rethink how operating systems and user interfaces are designed,” adding that the internet needs a protocol that is equally usable by people and agents. Two days later, OpenAI’s developer team shipped an open-source voice component that enabled a user to play chess on a webpage using only speech. No clicks. No keyboard.

The common thread is a bet that the basic unit of mobile computing shifts from the app (a container you navigate) to intent (a goal the system fulfills). If that bet is right, the intermediary between the user and the service shifts from the app store to an agent.

However, for this system to become viable, it will have to go head-to-head with the existing infrastructure, which is more than just a design choice.

What the app economy stands to lose

The current app model is a financial architecture where providers like Apple generated $109.34B in fiscal 2025, more than a quarter of total revenue. The system works on commissions and with over 850M weekly users, it facilitated $1.3T in developer billings and sales in 2024 alone.

Google’s Play Store operates on a similar model, and both companies derive strategic leverage from controlling what gets distributed and on what terms. The DOJ’s antitrust lawsuit against Apple, which survived a motion to dismiss in June 2025, is built on exactly this claim: that Apple uses its control over app distribution to block innovative services and lock in consumers.

An agent-first phone will be the most direct threat to this business model, as it will eliminate the mechanism for collecting commissions.

If a user never opens an app because the agent books the flight, orders the groceries, and schedules the appointment directly, the toll booth disappears. But so does the infrastructure developers currently depend on for distribution, discovery, and payment processing. The question is not just what the incumbents lose but what replaces the system for everyone else.

Companies like Apple and Google, which profit the most from the existing system, understand the risks and have been moving to ensure they have a say in the future of mobile devices.

Apple’s answer is already in motion.

At its WWDC 2025 event, Apple introduced the Foundation Models framework, which gives developers access to a ~3B-parameter on-device language model via native Swift code, with free, offline inference. A revamped Siri, powered by large language models, is expected later in 2026 and will be able to perform multi-step tasks via app intents.

Apple’s approach is to absorb the agent concept into its existing platform rather than replace the platform with the agent.

Beyond the technical shift, the company’s leadership transition further reinforced this idea. John Ternus, who takes over as CEO on September 1, is a mechanical engineer who joined Apple’s product design team in 2001 and oversaw hardware engineering for virtually every major product category. Bloomberg described him as someone who drills into tiny technical details in meetings, a different profile from Cook’s supply-chain orientation. A hardware-first CEO running a company whose AI strategy depends on on-device processing is a coherent combination.

Meanwhile, Google is embedding agentic capabilities directly into Android. Gemini is scheduled to replace Google Assistant fully in 2026, and its agentic features already perform multi-step tasks, such as ordering food and booking rides, autonomously on newer devices. Google does not need users to buy new phones, since it already has over a billion Android devices on which to deploy the agent layer via a software update.

The changing face of the smartphone

For OpenAI, a smartphone pushes its agenda and provides the impetus to transform the mobile device from an app-centric to an AI-based one.

The app grid has been the organizing metaphor of personal computing for 18 years. It may not survive the next five in its current form. But the question that lingers is whether the entity that sits between you and your intentions will be one you chose, one you can see through, and one that answers to you rather than to the company that built it. The app store, for all its faults, at least puts the icons where you can see them.

Headlines You Actually Need

• Surveillance state's expo moment: The Border Security Expo showed off how CBP, ICE, and DHS are quietly building one of the most advanced surveillance stacks ever deployed on American soil.

• Medicare's AI-ready overhaul: Medicare's new payment model is built from the ground up with AI, and most of the tech world hasn't yet noticed the massive opportunity.

• CEOs admit AI productivity is fuzzy: A growing number of CEOs are quietly acknowledging that AI productivity gains are harder to measure than the headlines suggest.

Friday Poll

Weekend To-Do

• Test an agent-first workflow: Try Perplexity's Comet browser or OpenAI's Operator for a real task, like booking travel or planning a meal, to feel what apps disappearing might actually look like.

• Audit your screen time by app: Use Screen Time on iOS or Digital Wellbeing on Android to see which apps an agent could realistically replace; you'll be surprised how few you actually need.

• Explore on-device AI: Download Apple's Foundation Models demo apps. The future of agentic phones runs locally, not in the cloud.

Meme Of The Day

The Toolkit

• Together AI: Cloud platform for running and fine-tuning open-source AI models at scale.

• VEED: Browser-based video editor with AI subtitles, voice cloning, and one-click translation.

• Writer: Enterprise AI writing platform with custom models trained on your brand voice.

Rate This Edition