AI on the Clock mehmeterkek.com

Pulse

5 August 2026

4 stories, chosen and edited by hand.

01

AI makes weather prediction better. Can WindBorne make it lucrative?

TechCrunch AI Tim Fernholz

WindBorne Systems raised a $37 million Series B co-led by Khosla Ventures and Galvanize, with TransLink Capital and Lux Capital participating, valuing the company at $250 million. Founded in 2019, it flies about 600 long-endurance weather balloons from 20 launch sites and feeds the data into its own AI forecasting model. Customers include the U.S. National Weather Service, Air Force, and Navy.

My readThe interesting part isn't the balloons, it's that AI weather models killed the supercomputer requirement, so a startup can now own the model and the data. But look at the customer list: National Weather Service, Air Force, Navy, plus hedge funds trading commodities. That's the same government-first trap earth observation startups fell into. Selling forecasts to a business that has no workflow for forecasts is still the hard part. Building the ship-onboard offline model, though — that I'd like to see.

Read the original at TechCrunch AI →

02

Anthropic signs $10B deal with AI cloud startup Volta

TechCrunch AI Lucas Ropek

Anthropic has reportedly signed a $10 billion, six-year compute deal with Volta, an AI cloud startup founded earlier this year, per Bloomberg. Volta is working with crypto-mining company Bitdeer on a 133 megawatt data center in Norway running Nvidia's Vera Rubin systems. Anthropic recently struck compute deals with SpaceX and Amazon too.

My readA company founded this year just landed a ten-billion-dollar contract, and the buildout partner mines crypto. That tells you how tight compute is right now — Anthropic will sign with almost anyone who can get megawatts and Vera Rubin racks online. Norway makes sense for power and cooling. What I'd want to know is delivery dates, because 133MW on paper and 133MW serving Claude traffic are very different things. Six years is a long time to find out.

Read the original at TechCrunch AI →

03

Apple says more ex-employees may have taken confidential data to OpenAI

TechCrunch AI Sarah Perez

Apple is seeking a preliminary injunction in its trade secrets case against OpenAI, aiming to block development of an AI device based on Apple technology, plus expedited discovery from OpenAI, its foundation, io, engineer Chang Liu and hardware chief Tang Yew Tan. Apple says 11 more former employees may be involved. OpenAI called the request "based on false information" and says it doesn't want Apple's trade secrets.

My readThe detail that stuck with me isn't the injunction, it's the part where multiple ex-Apple people at OpenAI reached out to hand back work laptops they'd kept. That's not espionage, that's offboarding. And OpenAI's counter that "residual access" was Apple's own bad security is the line every engineer reading this will nod at. Screenshots of unannounced products before an interview is a real problem. But a lot of this looks like a company discovering its own leaky exit process in public. I'm watching whether the injunction actually lands.

Read the original at TechCrunch AI →

04

Spotify expands AI remix and covers project with Merlin partnership

TechCrunch AI Sarah Perez

On Tuesday's Q2 earnings call, Spotify said Merlin, the licensing partner for independent labels and distributors, has joined Universal Music Group on its upcoming AI covers and remix product, bringing more than 30,000 labels. It'll start as a research preview for a subset of users, no date given, and launch as a paid add-on. Deezer recently said over 50% of daily track uploads are AI-generated, up from 10% in January 2025.

My readThe interesting bit isn't the AI, it's the plumbing: consent, credit, compensation, opt-in per artist. That's a rights ledger, and Spotify's bet is that owning it is the moat. Söderström saying "this product will not happen without us" is true and also an admission that plain generative music already happened without them. No date, no full catalog, a subset of users. I'd hold applause until I see what a remix actually pays.

Read the original at TechCrunch AI →