Elon is now after Github

Welcome back to The Shift. Let’s get straight to what matters in AI today…

Today we have:

💻 Cursor is moving onto Github’s turf

✨How to redesign any ui by describing it with the /design skill in Claude Cowork

⚡ OpenAI and Nvidia plan an 8 gw AI campus

🔨Tools and Shifts you cannot miss

💻 Cursor is moving onto Github’s turf

Cursor is expanding beyond being an AI coding environment with Origin, its own code hosting platform. The bigger play is bringing repositories, pull requests, reviews, and coding agents into one agent-native workflow.

The Shift:

1. Code Hosting Arrives - Developers can now create repositories and host code directly inside Cursor. Origin includes the fundamentals such as code browsing, search, pull requests, commits, checks, and merging.

2. GitHub Stays Connected - Teams don't need to migrate immediately. Existing GitHub repositories can be mirrored inside Origin in real time, with PR comments and activity syncing between both platforms.

3. Agents Join Repos - Cursor agents sit directly alongside the code and pull requests. They can explain code, make changes, update PRs, and push branches without developers constantly moving between separate tools.

4. GitHub Gets Competition - Origin launched on the same day GitHub suffered another major outage, with some functionality affected for over six hours. The timing highlights Cursor's larger ambition to own more of the development workflow rather than simply sit on top of existing infrastructure.

Cursor is evolving from an AI coding assistant into a development platform. If agent-driven software development keeps growing, controlling where agents write, review, host, and ship code could become far more valuable than controlling the editor alone.

Together with Harmonic Security

Your employees are connecting AI to everything. Now what?

ChatGPT and Claude don't just answer questions anymore. Employees are connecting them directly to Notion, Linear, Jira, and the rest of your stack. The AI can read, write, and take actions on company data. Most IT and security teams have no visibility into any of it.

Harmonic Security Connectors changes that. It sits inline with every AI-to-app connection, so you see each call, control what data moves, and block destructive actions before they happen. Employees notice nothing different.

See what's actually running across your business in a live demo.

See what's connected

✨How to redesign any ui by describing it with the /design skill in Claude Cowork

Anthropic's /design skill inside Claude Cowork turns a plain-language brief into multiple production-ready design variants, with real audit data, real files, and an interactive preview. Here's how to use it.

Step 1: Open Claude Cowork Launch Cowork from your Claude desktop app. This is where the /design skill lives and runs.

Step 2: Type /design and describe what you want Start your message with /design then describe the outcome. "Redesign the composer based on what people actually use it for." Claude pulls relevant data, checks your toolchain, and gets to work automatically.

Step 3: Let Claude audit and build Claude runs commands, pulls source data, creates files, and produces multiple distinct design variants, each with a different approach, rationale, and tradeoff explained in plain language.

Step 4: Review your variants Each option renders as an interactive preview in the right panel. Read the reasoning behind each one before picking. Version A might be conservative. Version B adaptive. Version C minimal. You decide what fits.

Step 5: Pick one and iterate Type which option you want implemented and describe any edits. "Ok, implement option B with my edits." Claude implements it, saves, and exports with no manual handoff needed.

From brief to interactive design variants in one command. No separate design tool. No back and forth with a designer. You can try it in Claude Cowork here.

⚡ OpenAI and Nvidia plan an 8 gw AI campus

OpenAI and Nvidia are taking the AI infrastructure race to another level with a huge new Ohio campus. The project combines Nvidia chips, enormous financing support, and nearly 8 GW of planned compute capacity.

The Shift:

1. Massive Ohio Build - The campus will rise on federal land around a former Cold War uranium facility in Pike County. Its first 800 MW phase is targeted for 2028, with the broader site eventually approaching 8 GW.

2. Nvidia Goes Bigger - Nvidia isn't simply supplying the chips. The company is also providing up to $105 billion in credit support and investing another $1.5 billion into SB Energy.

3. OpenAI Will Lease - SB Energy will develop and lease the campus to OpenAI. This structure gives OpenAI access to enormous compute capacity without requiring it to directly build and own the entire project.

4. Huang Answers Critics - Jensen Huang argues frontier AI labs are increasingly limited by infrastructure and financing rather than demand. Responding to accusations of circular financing, he was explicit: “No. OpenAI will pay the lease.”

AI competition is becoming an infrastructure and capital race as much as a model race. OpenAI's ability to secure nearly 8 GW of compute shows just how physical, expensive, and financially intertwined the next phase of AI could become.

Together with Guardian Financial Publishing

For AI Investors Who Do Their Homework Before They Buy.

Most people will chase the next AI IPO on hype alone and get burned. If you'd rather understand the timeline, the retail access window, and the risk buried in the filings before you commit a dollar, the free briefing is built for you.

Read the Free Briefing

🔨AI Tools for the Shift

🧑‍💻 Clears – Takes software work beyond AI coding by autonomously handling tasks from ticket triage through implementation and pull requests.

🌐 Pickle Browser – Gives AI agents a local browser they can operate visibly on your computer, with approval controls for important actions.

🧠 Zetik – Acts as an AI chief of staff in your pocket, helping organize and execute everyday work through autonomous agents.

🤖 Omni by xpander – Provides infrastructure for running AI agents without constantly monitoring, restarting, and manually managing their workflows.

🧪 Kubit – Uses real user behavior data to analyze and optimize the actions AI agents take inside products and business workflows.

🚀Quick Shifts

👀 Apple’s camera-equipped AirPods appear to have leaked. A macOS video appears to show AirPods using cameras and Visual Intelligence to understand surroundings, potentially giving Siri eyes into the real world.

📊 Similarweb can now track ads inside AI conversations. Its new AI Ads dataset tracks sponsored placements. Similarweb says ads appear in 26% of ChatGPT responses and nearly 30% of ad-eligible AI Mode queries. 

🏗️ Nvidia backs OpenAI’s massive Ohio data center. Nvidia will provide major credit support as OpenAI secures 8GW of capacity in Ohio, with the first 800MW expected online in 2028 

📈 Anthropic’s annualized revenue surges to $65b. Anthropic reportedly hit a $65B revenue run rate in July, up from $9B at 2025’s end, as it prepares for a potential record-breaking IPO. 

📚 Amazon is destroying rare books to train AI. Amazon is reportedly buying and dismantling rare books to scan their contents, giving its AI models access to valuable human-written text unavailable online. 

💃Prompt of the day

Create a Nostalgic Vacation Photo Grid

Image Prompt: Create a realistic 4x4 grid of 16 candid iPhone photos showing a young couple on vacation, photographed across spontaneous moments throughout their trip.

Context

• Subjects: [Young couple / upload reference images]

• Setting: [Insert vacation destination]

• Camera: Casual iPhone photography

• Mood: Nostalgic, intimate, playful, imperfect

• Layout: 4x4 photo grid

Instructions

Show the couple taking selfies together, photographing each other, laughing, walking, eating, exploring, relaxing, and sharing small unplanned moments throughout the vacation. Make every frame feel genuinely amateur and spontaneous with inconsistent angles, imperfect crops, occasional camera shake, slight motion blur, accidental overexposure, uneven lighting, tilted horizons, fingers near the lens, and subjects sometimes partially outside the frame. Vary the distance, poses, expressions, locations, outfits, and time of day while keeping both people visually consistent across all 16 images. Use subtle vintage color characteristics, soft highlights, natural smartphone noise, slightly faded tones, and authentic iPhone processing rather than polished professional photography.

🤳AI Shift of the Day

That’s all for today’s edition. See you tomorrow as we track down and get you all that matters in the daily AI Shift!

If you loved this edition, let us know how much:

Forward it to your pal to give them a daily dose of the shift so they can 👇