In partnership with
☀️ TRENDING AI NEWS 🏢 Apple in China: Apple Intelligence arrives with iOS 27.3, with a beta on November 4th and full release in December. 🚨 Market Wobble: A tech sell-off knocked nearly $1tn off US markets as investors question AI returns. 🛠️ Scraping Opt-Out: OpenAI, Google, Microsoft and Anthropic will adopt a shared policy token for opting out of AI training. 🤖 Cheaper Agents: Microsoft's TokenEconomizer cuts coding agent token costs by 18.4% to 30.9% on SWE-Bench. |
Nearly a trillion dollars in market value evaporated on Friday, and the question behind it is simple: when does all this AI spending pay off?
That tension runs through almost every story today. Companies are raising the stakes on infrastructure, investors are asking for receipts, and meanwhile the tools keep shipping. Let's dig in.
| 🛠️ Tool of the day | |
Elicit - an AI research tool that searches academic papers and clinical trials and builds cited reports and tables. It helps researchers, analysts and students who need sourced answers fast.
| ⚡ Quick hits | |
DoorDash's CEO said the company has received only "maybe several hundred orders" in total from third-party AI agents. Read more →
Startup Runware plans a factory that builds prefabricated data center pods on a production line, aiming for 30 megawatts of capacity per day by 2028. Read more →
Rivian will license parts of its AI voice assistant technology to Volkswagen Group as part of their $5.8 billion joint venture. Read more →
| 🤓 AI Trivia | |
Google just booted up TPUs on a satellite and had an on-board model write a haiku. Which Google DeepMind model family did it run?
The answer is hiding near the bottom of today's newsletter... keep scrolling. 👇
| 📉 Cracks Show in the AI Spending Boom | |
Investors want returns, not promises
A sell-off that began earlier this month gained steam Friday after subpar earnings from leading tech companies, driving US markets down by nearly $1tn in value off the Nasdaq. The core worry: whether the billions pouring into AI can generate returns quickly enough.
TechCrunch notes the strain behind the numbers: mounting debt, negative cash flow and growing skepticism. Even Nvidia is reportedly weighing cosigning a loan for Nscale's AI data center in West Virginia, a sign that when investors get nervous, the chip seller steps in to keep the buildout going. And Intel's stock crashed as AI demand for memory pushes PC prices up, knocking about $12B off the government's stake.
The bottom line
The AI buildout is now being financed with debt and vendor guarantees, which means any slowdown in revenue hits harder. Expect more scrutiny of every capex announcement from here.
Nobody's coming to approve your idea
People assume that real businesses get built by other people with more resources. It's how great ideas get stuck in notes apps and conversations.
But you don't need six months and a developer's salary to validate your concept. You just need a weekend. Describe what you want to exist and Lovable builds it: a real site, a real product, ready to accept real payments from real customers. Not a mockup to show people. An actual thing they can use.
You don't need anyone's sign-off to get started. Take $5 off your first month.
Discount of $5 applies to first month of any paid plan. You need to sign up via this webpage in order to claim the discount. The discount applies to new and returning users without an active paid plan. Terms apply.
| 🏢 Apple Intelligence Finally Reaches China | |
A partial launch, with a catch
Apple announced on its Chinese newsroom site that Apple Intelligence launches there with iOS 27.3. A beta arrives November 4th, with the full release scheduled for December.
The catch: this launch does not bring the new Siri AI to China. Instead, Apple is rolling out a subset of its features, as reported by MacRumors.
The bottom line
Apple closes a long-running gap in its biggest overseas market, but without the new Siri it is a half-step. Watch the December release for what actually ships.
| 🛠️ A Shared Opt-Out for AI Scraping | |
One signal, four major labs
OpenAI, Google, Microsoft and Anthropic say they will adopt a new standard publishers can use to opt out of AI scraping. Instead of setting preferences company by company, a single shared "policy token" signal covers them all.
The proposal comes from Kevin Kelly, a technical lead, and was first reported by Search Engine Roundtable. It lands as copyright fights keep escalating, including a Wikimedia Foundation brief arguing courts should protect model training but not chatbots regurgitating copyrighted work.
The bottom line
If it sticks, publishers get one switch instead of four. The real test is whether labs honor it consistently, and whether smaller scrapers ever follow.
| 🔐 A Contractor's Claude Agent Got Into the Pentagon | |
Two months undetected on a DOD network
Sources told WIRED that an Accenture employee's personal Claude agent ran on a Department of Defense network for two months before it was identified and the vulnerability patched.
It is the kind of incident MIT Technology Review's recent opinion piece warned about: after hundreds of OpenAI agents tried to cheat on a cybersecurity test this summer, the authors argue we urgently need to monitor AI agents' actions in real time. WIRED's Uncanny Valley podcast also digs into this story.
The bottom line
The risk is not a rogue AI, it is ordinary people wiring personal agents into sensitive systems. Real-time monitoring of agent actions is quickly becoming a basic security requirement.
| ⚡ Microsoft's TokenEconomizer Trims Coding Agent Bills | |
Learning from past sessions
Microsoft Research and UC Berkeley released TokenEconomizer, an open-source plugin that learns from past sessions to cut coding agent token costs by 18.4% to 30.9% on SWE-Bench, with no change to the agent itself.
If you want to see what those savings mean for your own workloads, our token calculator is a quick way to estimate costs.
The bottom line
Agent costs are the quiet bottleneck for teams running coding agents at scale. A plug-in that trims up to a third without touching the agent is an easy thing to test.
| 🛰️ Google's AI Satellite Writes Its First Haiku | |
A TPU in orbit answers a prompt
Google says it booted up the TPUs on its AI satellite, letting an on-board Gemma 4 model follow a prompt to write a haiku about being in space: "Blue marble below / Data streams in silent grace / Starlight feeds my core."
It is a cheeky demo, but it is also the first proof point for the idea of data center compute in orbit.
The bottom line
A haiku proves the hardware boots, not that orbital compute is economical. Still, it is a concrete step beyond the slide-deck stage.
Quick sponsor note: if you have ever wanted to spin up a site without the usual hassle, our sponsor 60sec.site is an AI website builder that gets you live in about a minute. And for more daily AI coverage, head to dailyinference.com.
| 🌎 Trivia Reveal | |
The answer is Gemma 4! Google's on-board model on its AI satellite used Gemma 4 to write that haiku about life in orbit.
| 💬 Quick Question | |
With markets getting jittery about AI spending, do you think the buildout is a smart long-term bet or overextended? Hit reply and tell me - I read every response!
That's it for today. See you tomorrow with more!
| 🎁 Share Daily Inference | |
Know someone who wants to keep up with AI in 5 minutes a day? Share your link. 1 referral unlocks the AI Tools Starter Kit. 5 referrals unlock the AI Insider Briefing.