☀️ TRENDING AI NEWS
🤖 OpenAI shuts down Sora app just six months after its splashy launch
🏢 Arm unveils its first ever in-house CPU, with Meta as launch customer
⚖️ Federal judge questions Pentagon's "attempt to cripple" Anthropic in court
🛠️ Anthropic gives Claude Code autonomous computer control with new auto mode
Three stories today. Three very different companies. And somehow they all point at the same thing: the AI industry is entering a phase where the hype has to start meeting reality.
🤓 AI Trivia
How long has Arm Holdings been in the chip design business before releasing its very first in-house CPU this week?
⏳ 15 years
⏳ 20 years
⏳ 35 years
⏳ 45 years
The answer is hiding near the bottom of today's newsletter... keep scrolling. 👇
🤖 OpenAI's Sora Is Dead
Six months. That's how long OpenAI's TikTok-style Sora video app lasted before the company quietly pulled the plug. The announcement came Tuesday afternoon - a brief "We're saying goodbye to Sora" with no ceremony and, frankly, no surprise from anyone who'd used it.
From Disney Deal to Digital Dustbin
What makes this particularly awkward is the timing. Just a few months ago, OpenAI centred Sora in a massive licensing deal with Disney. Now that deal appears to be dead too, according to The Verge. The standalone app - a scrollable social feed of AI-generated videos - never found a sustainable audience. TechCrunch noted there simply wasn't "sustained interest" in an AI-only social feed, which honestly tracks.
The underlying Sora 2 video generation model isn't going anywhere - it'll remain accessible through OpenAI's API and ChatGPT. But the bet on a consumer app built around AI video slop? That experiment is over.
🏢 Arm Breaks a 35-Year Rule
For its entire 35-year existence, Arm has been the company that designs chip blueprints and lets everyone else build them. Qualcomm, Apple, Samsung - they all license Arm's instruction set architecture and produce their own silicon. That was the deal. This week, Arm changed it.
Meta Gets the First Delivery
The new chip is called the Arm AGI CPU, and it's purpose-built for AI inference workloads - specifically the kind of compute needed to run AI agents that spawn cascading sub-tasks in the cloud. Meta is the debut customer, with deliveries expected later this year. OpenAI, Cerebras, and Cloudflare are also in line according to Wired.
The obvious tension here: Arm's entire business model is built on licensing to the same companies now watching their supplier become a competitor. CEO Rene Haas is arguing the market needs this chip and that it won't undercut existing licensees - but that's a difficult line to walk. The semiconductor industry is watching closely.
⚖️ A Judge Just Put the Pentagon on the Defensive
The Anthropic vs. Department of Defense case got significantly more interesting yesterday. During a federal court hearing, a district judge openly questioned whether the Pentagon's decision to label Anthropic a "supply-chain risk" was a genuine security concern or - and this is the judge's language - an "attempt to cripple" the company.
Refusing Weapons Work Has Consequences
The backstory: Anthropic refused to allow its Claude model to be used in autonomous weapons systems. The Trump administration then ordered US agencies to stop using Anthropic's AI entirely, citing supply-chain risk. Anthropic is seeking a temporary pause on that order while the case plays out.
MIT Technology Review's AI Hype Index noted that while Anthropic drew the weapons line, OpenAI moved quickly to secure Pentagon contracts in what analysts described as "opportunistic" timing. The military AI question isn't going away - and this court case may end up setting a precedent for how AI companies handle government pressure on what their models can and can't do.
🛠️ Claude Code Gets the Wheel
While Anthropic is fighting in court, its engineering team shipped something genuinely useful. Claude Code now has an "auto mode" that lets the AI execute tasks on your computer without asking for approval at every step. It can open files, browse the web, run dev tools, and operate while you're away from your machine.
Autonomy With a Built-In Brake Pedal
The key word in Anthropic's framing is "safeguards" - the company is pitching this as a balance between speed and safety. It requires no setup, which is notable for developer tooling. This fits a broader industry trend toward agentic coding tools that operate more like junior developers than autocomplete engines.
If you're building apps and haven't set up a local AI workflow yet, tools like this are moving fast. Speaking of building fast - 60sec.site is an AI-powered website builder that gets you from idea to live site in under a minute. Worth a look if you're spinning up landing pages or prototypes.
💰 $3.5 Billion Says AI Isn't Slowing Down
Kleiner Perkins just closed a $3.5 billion fund with an explicit AI focus - $1 billion earmarked for early-stage startups and $2.5 billion for late-stage growth companies. This is one of the largest fundraises in the firm's history.
It's a statement fund. After years of debate about whether AI valuations are sustainable, one of Silicon Valley's most storied venture capital firms is doubling down hard. The split between early and growth stage is interesting too - it signals they want exposure across the full AI company lifecycle, not just seed bets.
🌎 Trivia Reveal
The answer is 35 years! Arm was founded in 1990 and spent its entire history as a chip design licensing business - selling blueprints, never building. This week's Arm AGI CPU is the first time the company has produced its own silicon, making Meta its first-ever direct chip customer.
💬 Quick Question
The Sora shutdown is a rare "product fails and dies" moment in a space where companies rarely admit defeat publicly. Are there other AI products you think are living on borrowed time right now? Hit reply and tell me what you'd put on the chopping block - I read every response!
That's all for today. Stay up to date with everything AI at Daily Inference - see you tomorrow!