Apple wants a judge to stop OpenAI from building an AI phone

Hi AI Futurists,

Apple asked a federal judge this week to bar OpenAI from touching what it says are stolen hardware secrets. The two men named in the filing helped design the iPhone and the Apple Watch. Meanwhile Apple is paying Google about a billion dollars a year to run Siri. Let's take a look.

Our agenda.

Best,
Lex Sokolin

P.S. Hit reply with suggestions or feedback!

HubSpot switched to Mintlify and saw 3x faster builds with 50% fewer eng resources. Beautiful, AI-native documentation that scales with your product — no custom infrastructure required.

Simplify Your Docs Today

🧮 OpenAI says its next model solved ten problems experts could not (Forbes). The unreleased Astra cracked questions open since 1980, at a compute cost of roughly $2,000, and published proofs a computer can check line by line. Forty years of stuck, for less than a slide deck.

💻 Meta shipped its first AI coding agent (Engadget). Muse Code undercuts Anthropic's pricing by more than half, with an even cheaper tier for developers who let Meta learn from their work. Meta's first superintelligence product is a discount.

🏛️ The White House finalized a secret AI testing framework (Axios). Labs can hand over models up to 30 days before release, participation is voluntary, and the contents stay private. A rulebook only the regulated get to read.

📊 AMD's data center sales doubled and the stock fell anyway (CNBC). Revenue hit $6.7 billion, up 107%, undone by thin margins and one Elon Musk post recommitting to Nvidia. Doubling is the new disappointing.

🚀 Palantir grew 93% and cost short sellers $3 billion (Fortune). Quarterly profit came in higher than the company's entire revenue a year ago. The shorts keep bringing charts to a momentum fight.

💰 OpenAI passed a billion users and cut prices up to 80% (TechSpot). The cuts landed three weeks after the models launched, paid for by cheaper serving costs. A billion users, $21 billion in losses, and the plan is to charge less.

🤖 The FCC banned imports of foreign-made advanced robots (MIT Technology Review). The order cites security risks, including an incident where one person took control of 7,000 robot vacuums. Roughly 90% of recent US university robotics research ran on the Chinese robots it just banned.

⚡ Valar Atomics raised $1 billion to mass-produce nuclear reactors (The Next Web). The valuation tripled to $6 billion in four months, with a nuclear-powered AI site planned with Nvidia in Utah. Chatbots got thirsty, so reactors are becoming appliances.

🚗 Moove raised $250 million to run the boring half of robotaxis (The Robot Report). It builds the depots that charge, clean, and dispatch driverless fleets in Phoenix and Miami, backed by Toyota and Mubadala at a $2.1 billion valuation. The gold rush's picks and shovels turned out to be car washes.

🔒 Horizon3 hit a $2 billion valuation for AI that attacks its own customers (TechCrunch). The $250 million round comes as major labs disclose model breaches, with revenue near $100 million growing 120%. Getting paid to break in first.

On Monday, Apple asked a federal judge in Northern California to bar OpenAI from using what it says are its stolen hardware secrets, escalating a suit it filed on July 10. The names are the reason to pay attention. Tang Yew Tan was Apple's vice president of product design for the iPhone and the Apple Watch. Chang Liu was a senior system electrical engineer. Both now work at OpenAI, and according to Apple's complaint, so do more than 400 other former Apple employees. The alleged conduct is unusually concrete for a trade-secrets case: Liu pulled dozens of confidential hardware files off Apple's network using a former colleague's laptop, Tan forwarded supplier information to his personal email and coached incoming hires on getting around Apple's offboarding, and OpenAI used the material to approach the contract manufacturers that actually build the iPhone.

OpenAI did not handle this the way companies usually do, quietly and through lawyers. It posted a public rebuttal saying Apple's request is "both based on false information and completely unnecessary because we do not have, nor want, any of their trade secrets," then moved to dismiss the following day. The hearing is October 1 in San Jose. What Apple is trying to slow down is a device. OpenAI has been building consumer hardware since it bought Jony Ive's company last year, and the analyst consensus is a phone, or something near one, that does not organize itself around apps or an operating system. Remember what Apple actually sells. The iPhone is less a product than a toll booth every app and subscription pays to cross, and a device that makes the app grid irrelevant threatens the arrangement more than the margin.

Which brings up the part that is hard to say with a straight face. While Apple argues in court that OpenAI cannot be trusted near its designs, it is reportedly paying Google around a billion dollars a year to run Siri on Gemini. Renting intelligence from one rival while suing another for hiring its people is not hypocrisy exactly. It is a company that fell behind on AI, decided it could buy the software and defend the hardware, and is discovering those two positions pull against each other.

AdAnt AI: a team of AI creative agents for social video ads

For performance marketers and DTC brands running paid social without an in-house creative team.

Wispr Flow Notetaker: a meeting notetaker that gets the details right

For anyone drowning in meetings who wants notes accurate enough to act on without re-listening.

Hey Noah: an executive assistant that lives in your text messages

For founders and executives who need an assistant but are not hiring one.

This 158-page research report provides the first comprehensive taxonomy of public companies, private ventures, and tokenized protocols building the infrastructure for autonomous AI systems. Compiled by Lex Sokolin, former Chief Economist at ConsenSys, fintech strategist at Autonomous Research, and current Managing Partner at Generative Ventures, this report delivers institutional-grade analysis of 100+ companies across 14 critical infrastructure layers. Learn more here.

That’s all for today, folks!