In partnership with
☀️ TRENDING AI NEWS 🚨 Market Rout: The Nasdaq tumbled 5.5% on Thursday, its worst day in 19 months, shaving nearly $2tn off its value. 🤖 Open Model: Thinking Machines released Inkling 2 Small, a 594B MoE with just 17B active parameters. 🏢 Acquisition: Palo Alto Networks is in talks to buy AI agent security startup Clutch for $500M. 🛠️ Claude Update: Claude's new Personal Finance feature connects to your bank and card accounts through Plaid. |
Nearly $2 trillion vanished from the Nasdaq in a single session, and the trigger was not a chip shortage or a bad earnings call. It was an election.
Wall Street's AI trade just collided with politics, and the fallout reached Asia and Europe by Friday morning. Today we cover the sell-off, a very fast pivot from OpenAI, and a big new open model.
| 🛠️ Tool of the day | |
Fathom - records, transcribes and summarizes your Zoom, Google Meet and Microsoft Teams calls, so you can skip the note-taking. It helps anyone who spends the day in meetings. It has a free plan.
| ⚡ Quick hits | |
Ricursive Intelligence released Ripple-65B Preview, a 65B open-weight model under Apache 2.0 that the company says leads U.S. open models on two coding benchmarks. Read more →
| 🤓 AI Trivia | |
In a mixture-of-experts (MoE) model like Inkling 2 Small, what does 'active parameters' refer to?
The answer is hiding near the bottom of today's newsletter... keep scrolling. 👇
| 📉 Nasdaq Suffers Its Worst Day in 19 Months | |
Midterm results spook the AI trade
The Nasdaq composite slumped 5.5% on Thursday, tumbling into correction territory as investors bet that the Democrats' midterm victory means tougher regulation for the AI industry. Nearly $2tn was shaved off the index, and worries about curbs on AI datacenters rippled across bourses worldwide.
By Friday, Asian and European markets followed, with the FTSE down 1.5%. To keep this in perspective, the Nasdaq is still up 4.6% for the year, though it now sits more than 10% below its peak. We also covered the data center backlash at the ballot box in yesterday's issue, and the market is now pricing it in.
The bottom line
This is the first time AI regulation risk has moved a major index this hard. Expect more volatility every time policy headlines hit, and expect AI companies to get a lot more careful about how they talk to communities.
Vanity Fair’s New Show, Where Nothing Is Off the Table
On Fair Game from Vanity Fair, join Mark Guiducci for in-depth and far-reaching conversations with the most interesting characters from the worlds of Hollywood, Politics, Style, Silicon Valley, and beyond.
From acclaimed actors and political figures to tech leaders and style visionaries, Fair Game goes beyond the headlines to uncover the people, personalities, and ideas driving the cultural conversation.
| 🏢 OpenAI Steps Back From NDAs on Data Centers | |
A fast response to voter anger
Following a midterm week where voters loudly opposed data center expansion, OpenAI announced new commitments to communities hosting its projects. The headline pledge: no more NDAs with public officials. Many data center projects have drawn scrutiny for using non-disclosure agreements to hide negotiations between developers and local governments.
Meta is sounding a similar alarm from the other direction, warning that growing political opposition could make future AI buildouts harder. Meanwhile, OpenAI president Greg Brockman said the company could be IPO-ready in 2027, which makes public goodwill a bigger deal than ever.
The bottom line
Secrecy was the cheapest way to move fast on data centers, and voters just made it expensive. Watch whether other labs and cloud providers follow OpenAI's lead on transparency.
| 🤖 Thinking Machines Releases Inkling 2 Small | |
594B parameters, 17B doing the work
Thinking Machines Lab released Inkling 2 Small, a mixture-of-experts model with 594B total parameters and only 17B active per token. It scores 31 on the Artificial Analysis Intelligence Index and runs on the company's Tinker platform, with Apache 2.0 weights planned.
That sparse design is the point: you get the capacity of a huge model while paying inference costs closer to a much smaller one. If the Apache 2.0 weights land as promised, it becomes another serious entry in the open-source AI race. If you want to estimate what running models like this would cost, our Token Calculator is a handy starting point.
The bottom line
Weights are not out yet, so treat this as a strong promise rather than a download link. If they ship under Apache 2.0, it is worth benchmarking against your current setup.
| 🛠️ Claude Plugs Into Your Bank Account | |
Personal Finance arrives via Plaid
Anthropic launched a Personal Finance feature for Claude that connects to your bank and card accounts through Plaid. Ask where your money went this month and it builds an overview with spending broken down by category, such as rent and utilities, shopping, and food.
It is a notable move into everyday consumer territory, and it lands the same week Wired reported that Anthropic is holding its vow to keep ads out of Claude while the rest of the chatbot industry races toward advertising.
The bottom line
Handing a chatbot read access to your accounts is a real trust decision. The convenience is clear, so the question is whether you are comfortable with a third-party bridge like Plaid in the middle.
| 🔐 Palo Alto Networks Eyes a $500M AI Agent Security Deal | |
Palo Alto Networks is in talks to acquire Clutch, an AI agent security startup backed by Lightspeed and SignalFire, for $500M. Clutch was last valued at $85 million, according to a source, which would be a jump of nearly 6x.
The price tag shows how quickly AI agents are turning into a security category of their own. Companies deploying agents that touch real systems need ways to monitor and constrain them, and incumbents are paying up to get there first.
| 🏠 Airbnb Is Building an AI Travel Agent | |
Airbnb said in its Q3 shareholder letter that it will roll out an AI agent in the first half of 2027 to help users plan and book travel. The company argued that today's agents handle simple tasks like restaurant reservations well, but planning a trip is a different, messier problem. The agent is meant to help compare options, manage bookings, and assist during a stay.
A reality check from Uber this week: during its busiest week last quarter, only four people booked a ride through an AI agent. Demand for agent-driven booking is still more promise than habit.
| 🌎 Trivia Reveal | |
The answer is: the parameters used to process each token! In a mixture-of-experts model, a router sends each token to a small subset of experts. That is how Inkling 2 Small can hold 594B parameters in total while using only 17B for any given token, keeping inference far cheaper than a dense model of the same size.
| 💬 Quick Question | |
Would you let an AI assistant like Claude read your bank and card transactions in exchange for instant spending breakdowns? Hit reply and tell me where you land - I read every response!
Quick plug: if you need a website fast, our sponsor 60sec.site is an AI website builder that gets you live in about a minute. And you can always catch our daily AI roundup at dailyinference.com.
That's it for today - enjoy your Saturday, and see you tomorrow with more.
| 🎁 Share Daily Inference | |
Know someone who would enjoy this? Share your link below. One referral unlocks the AI Tools Starter Kit, and five unlock the AI Insider Briefing.