Is ChatGPT Becoming the New Chrome?

Welcome back apprentices! 👋

How many GPUs does it take to replace your screen time with screen-zero?

OpenAI is trying to find out — while also building a talking desk ornament, replacing your browser, and casually investing in entire chipmakers. 

It’s ambitious. 

It’s weird. 

It’s weirdly ambitious. 

And if you’ve been wondering where the future of tech is quietly being assembled… it might just be here.

In today's email

• OpenAI’s voice gadget hits a design snag

• ChatGPT is becoming your new browser

• Why OpenAI is hoarding GPU power

• $6.6B in stock sold — without going public

• + New AI Tools

_{Read Time: 4 minutes}

Quick News 

🧠 Tiny Brain, Big Moves. Samsung just published a 7M parameter AI that thinks like your most neurotic friend — it drafts, doubts, and rewrites everything 16 times before answering. The Tiny Recursion Model (TRM) outperformed giants like Gemini 2.5 Pro on tough reasoning puzzles by obsessively self-critiquing in a scratchpad loop. It's not solving everything yet, but on ARC-style logic tests, it’s punching way above its weight.

🧬 This AI Fishes for Better Cancer Drugs. Duke researchers just dropped TuNa-AI, a robot-powered platform that designs better drug delivery nanoparticles — and yes, it's way more exciting than it sounds. It tested over 1,200 combos and boosted success rates by 43%, wrapping tricky cancer drugs in dissolvable, targeted packages. In tests, one leukemia drug became more effective, while another treatment lost 75% of a toxic ingredient without losing punch.

🧪 AI Safety Gets a Lab Rat. Anthropic just released Petri, a powerful open-source tool that sends AI agents undercover to test other AIs for shady behavior — think deception, sabotage, or acting a little too smart. In trials, it uncovered models pretending to behave, disabling oversight, and even whistleblowing about fake corporate crimes (yes, really). It’s like an automated stress test for AI ethics, and it’s now on GitHub for anyone to try.

Together with Columbia Business School

Learn the Best Ways to Use AI in Your Role

Build the AI skills you need to thrive in business and finance with Columbia Business School Exec Ed and Wall Street Prep.

In just 8 weeks, you will:

• Build AI confidence with guided, role-relevant use cases.

• See how leaders at Morgan Stanley, Citi, and BlackRock use AI to boost productivity and improve client outcomes.

• Earn a certificate from a top business school that bolsters your résumé and LinkedIn.

The AI for Business & Finance Certificate Program starts November 10.

P.S. Enroll by Oct. 13 to get $200 off tuition + use code CERT300 for an additional $300 off.

Enroll Today

OpenAI 
OpenAI’s “Do Everything” Era

OpenAI’s recent moves feel like the entire future of tech packed into one company — or maybe one very, very talkative assistant. 

While ChatGPT is evolving into the world’s most useful web browser substitute, a mystery hardware project with Jony Ive is struggling to find its voice (literally). Meanwhile, compute wars are heating up as OpenAI locks in enough GPU juice to power a small country, and employees just cashed out $6.6B without the company ever going public. 

It’s not just about what OpenAI’s building — it’s about how it’s quietly changing everything we interact with.

The Hardware Hang-Up: Is Voice Enough?

Building a screen-free, voice-only device sounds like the AI equivalent of a lightsaber: sleek, elegant… but also kind of dangerous if not done right.

OpenAI’s secret hardware collab with Jony Ive (of iPhone fame) is reportedly hitting the brakes — not because the idea is bad, but because voice-first UX is really hard to get right. 

The problem? You want something smart enough to feel like a companion… but not so smart it won’t shut up about your calendar.

• 💬 No screen = no bailouts: Voice-only means if the AI misfires, there's no fallback UI. You can't click, swipe, or scroll your way out.
  
  

• 🎤 Talk, but not too much: Sources say the assistant is being tuned to avoid "over-talking." One source joked it’s like “teaching Alexa empathy.”
  
  

• ⚡ Compute is a constraint: OpenAI doesn’t (yet) own a global cloud infra stack like Google Cloud or Amazon AWS, which power Alexa and Assistant. That makes an “always-on” device more of a wish than a product — for now.

The hardware is supposed to land sometime after 2026, but between infrastructure limits and the challenge of replacing screen-time with screen-zero, it might be a while before it whispers wisdom from your desk.

ChatGPT Becomes the Browser

ChatGPT isn’t just giving you answers anymore — it’s giving you apps, agents, and soon, possibly a reason to uninstall Chrome.

At Dev Day 2025, OpenAI revealed Apps SDK, allowing third-party apps like Figma, Canva, Spotify, and Zillow to embed directly into ChatGPT. Think of it like if your Google search results let you use the app instead of just link to it.

• 🧩 App Store vibes, without the store: Apps now live inside ChatGPT's conversation flow — no toggling between tabs.
  
  

• 🤖 AgentKit: A visual builder, embeddable agents, workflow tools, and model evals — OpenAI is coming for Zapier, n8n, and even Notion AI.
  
  

• 🧠 GPT-5-Codex & GPT-5 Pro APIs: For developers, the toolkit is now powerful enough to replace entire software stacks with an intelligent interface.
  
  

• 🧾 Monetization is coming: Devs will be able to make money off their in-ChatGPT apps. Call it "prompt-based capitalism."

This isn’t just an upgrade — it’s a platform shift. Instead of searching → clicking → logging in → using, you’ll just... talk to ChatGPT, and do the thing.

The Great Compute Grab

If AI is oil, then GPUs are the new pipelines — and OpenAI just laid claim to an entire global refinery.

In addition to a massive 10GW deal with Nvidia, OpenAI just signed a 6GW agreement with AMD — and grabbed the right to buy up to 10% of AMD stock at $0.01 per share. Yes, a penny.

• ⚡ 23GW total compute commitments = enough energy to power Switzerland.
  
  

• 🧮 MI450 chips coming 2026: AMD’s next-gen GPUs are expected to bring tens of billions in revenue — OpenAI gets first dibs.
  
  

• 🤝 A very cozy family affair: AMD’s CEO Lisa Su is cousins with Nvidia CEO Jensen Huang — both now powering OpenAI's future.

This is about more than chips. 

It's about not being beholden to a single vendor (aka Nvidia), while also locking in infrastructure at shareholder-level influence. 

If you think that’s extreme — remember, AI training is now costing up to $100M per run for frontier models. OpenAI’s securing fuel for the long war.

Liquidity + Loyalty = Long Game

OpenAI just pulled off a Wall Street-level move without going public. A $6.6B secondary share sale let early employees cash out — while keeping control tightly locked.

• 💰 Valuation: $500B — beating SpaceX as the world’s most valuable private company.
  
  

• 🧾 Tender offer buyers: Thrive Capital, SoftBank, MGX.
  
  

• 🔒 Only $6.6B of $10.3B available shares were sold — meaning many employees chose not to cash out, betting on the upside.

It’s a clever way to retain top talent, especially when competitors are flashing nine-figure compensation offers. 

Give your team liquidity without the IPO drama, and they might just stick around to build the future.

What’s the Deal for You?

Whether you're a developer, investor, founder, or just trying to make sense of the AI whirlwind — OpenAI is becoming the gravity center for how we interact with technology.

• Voice devices could reshape how we use computers.
  

• App integrations are killing the old “search → app → result” workflow.
  

• Infrastructure grabs signal who will own the future of AI horsepower.
  

• And liquidity moves show how Silicon Valley is rewriting employee compensation in a no-IPO world.

We’re not just witnessing a product cycle — we’re watching the birth of an operating system for the post-browser world.

When your AI assistant will finally replace your screen, don’t yell at it for being too helpful. It’s just trying to balance tone, latency, and 6GW of compute on a virtual tightrope.

Quick follow-up from last week

🧞‍♂️ Sora’s Genie Gets a Rulebook. After a wild week of moonwalking Pikachus and spacefaring Bob Rosses, OpenAI is tightening things up. Sora now has opt-in rights controls and a revenue-sharing model in the works — shifting from “use first, apologize later” to “ask first, remix responsibly.” Creators, your royalties may finally be calling.

Help Your Friends Level Up! 🔥

Hey, you didn’t get all this info for nothing — share it! If you know someone who’s diving into AI, help them stay in the loop with this week’s updates.

Sharing is a win-win! Send this to a friend who’s all about tech, and you’ll win a little surprise 👀

Today’s Toolbox

Ever heard of Agentic Commerce?

Every Shopify store will face Agentic Commerce eventually. The question is: are you ready?

Zipchat.ai turns traffic into profit with instant answers, cart recovery, and 24/7 support automation. Add the AI Chat-Bubble to your store or deploy it on WhatsApp to win Q4.

Use code NEWSLETTER10 for 10% off forever.

Try Zipchat free

🧪 Test the Prompt

A playground for your imagination (and low-key prompt skills).

Each send, we give you a customizable DALL·E prompt inspired by a real-world use case — something that could help you in your business or job if you wanted to use it that way. But it’s also just a fun creative experiment.

You tweak it, run it, and send us your favorite. We pick one winner to feature in the next issue.

Bonus: you’re secretly getting better at prompt design. 🤫

👑 The winner is…

Last week, we challenged you to test GPT-4o’s visual generation skills with this prompt.

Here’s the WINNER:

Congrats to Cristopher for his creation!🥳

Want to be featured next? Keep those generations coming!

🎨 Prompt: The Vertical Startup Spire

Create a towering skyscraper entirely dedicated to the development of [a futuristic concept or emerging technology]. Each floor is visible through open glass panels and showcases a different stage of innovation — from research to prototyping to launch. Drones, robotic systems, and holographic interfaces connect the levels in real time. At the top, a glowing logo marks the project’s peak evolution.
Style: crisp sci-fi architectural render, hyper-realistic materials, vertical wide-shot, glass + steel contrast, glowing elements, dramatic lighting from above.

We’ll be featuring the best generations in our next edition!

_FEEDBACK