😸GPT-5 Codex codes for 7 hours straight?!

PLUS: AI bubble = good news?? 😵‍💫 🤑

Welcome, humans.

AI adoption is outpacing every technology in history, but this speed is creating massive inequality gaps. Anthropic just released fascinating data showing that wealthy countries like Singapore and Israel are using Claude 4-7x more per capita than emerging economies.

Even within the US, Washington DC and Utah are crushing California in per-capita usage. Countries with lower adoption rates use AI almost exclusively for coding, while high-adoption areas spread usage across education, science, and business.

See, the data reveals something striking: as AI capabilities improve, users in wealthy regions collaborate more with AI, while users in emerging economies just delegate complete tasks to it. In a way, we're watching different parts of the world develop entirely different relationships with AI technology.

But is that the whole story? Perhaps not. We put together our key insights from Anthropic’s data and OpenAI’s new landmark study on ChatGPT usage data into a deep dive comparing and contrasting the findings from both reports. Check it out here.

Here’s what happened in AI today:

OpenAI released GPT-5 Codex, a special model for using its code agent Codex.
The AI bubble is real, but OpenAI’s chair says that's actually good news.
Fiverr to layoff 30% of company as many tech firms consider restructuring.
Religious chatbots are low key crushing it w/ 30M downloads and counting.

Advertise in The Neuron here

OpenAI Board Chairman Says We're Living Through AI's Dot-com Moment (And That's Actually Good)

Remember when everyone called the dot-com boom a bubble, then watched Amazon and Google become trillion-dollar companies? OpenAI board chairman Bret Taylor thinks we're living through AI's version of 1999, and that's not necessarily bad news.

In a recent interview at The Verge, Taylor echoed Sam Altman's prediction that “someone is going to lose a phenomenal amount of money” in AI, but he's not sweating it. His take? We're absolutely in a bubble, but bubbles can birth the next generation of tech giants.

Here's his logic: The dot-com crash killed companies like Pets.com. But it also gave us Amazon, Google, and pretty much every major tech company that runs our world today. Even “failed” ideas later became successful companies like Instacart and DoorDash.

But Reddit users aren't buying it, saying that AI actually shouldn’t be compared to the dot-com bubble. The post pointed out that Pets.com had just $600K in revenue before imploding, while OpenAI just announced $10B in annual revenue. AI is already embedded across enterprise workflows, government systems, and education.

Without question, AI is genuinely being used across industries (just look at the Anthropic data in today’s blurb), but the profitability question remains thorny. And here's where it gets tricky. Despite massive revenues, major AI companies are still spending enormous amounts on compute infrastructure, top talent, and free user services. The costs of running these models at scale remain a significant challenge.

The key difference this time: Unlike the dot-com era where companies could burn cash for years chasing growth, today's AI boom demands faster results. Companies need to prove ROI quickly or risk getting cut off.

Here’s what this means for AI companies:

Some will become the next Amazon (massive winners).
Others will become the next Pets.com (spectacular failures).
The underlying AI infrastructure will get used, even if early builders go bankrupt.
Companies that can't show quick value will get weeded out faster.

Whether you buy the bubble comparison or not, one thing's clear: we're in a period where massive money is chasing AI opportunities. The companies that survive will be those building real solutions with clear paths to profitability, not just impressive demos.

Case in point: OpenAI just launched GPT-5-Codex, a specialized model built specifically for coding tasks that can work autonomously for over 7 hours on complex projects (and Every says you can get it to code for 35 minute straight “if you ask it nicely.”).

Here’s a few more details:

90% faster task completion through improved cloud infrastructure.
Automatic code reviews that catch critical bugs other tools miss.
Dynamic thinking that's snappy for simple tasks but goes deep on complex code refactors (restructuring messy code to make it better without changing what it does).

The proof it's working? Codex usage grew 10x in the past month, with companies like Duolingo, Virgin Atlantic, and Cisco Meraki using it for real production work.

“Codex caught a real active bug that other code review tools missed,” says Austin Ray from Ramp. “Precision and value are high per caught bug.”

This is exactly what Taylor meant about surviving the bubble. OpenAI is building tools that developers actually pay for and use daily. While tools like “Agent” still feel more in the flashy demo stage, Codex is real…and its slowly becoming the tool of choice over Anthropic’s Claude Code (although Anthropic’s data shows Claude Code usage is showing no sign of slowing down just yet).

Whether we're in a bubble or not, whether the current pricing is sustainable or not, companies showing this kind of concrete business value are probably going to be fine.

FROM OUR PARTNERS

Saddle Up: Warp launches Warp Code

Building software with AI can feel like wrangling cattle: agents get stuck going in circles or bolt in the wrong direction.

Warp Code is here to change that.

Warp is the best way to code with AI agents from prompt to production. With Warp Code, new features like a built-in editor and code review join existing favorites like MCP and CLI tools, all in one place.

Why Warp?

Top-rated agent: #1 on Terminal-bench, top-3 on SWE-bench Verified
All the top models, one subscription: Opus, Sonnet, GPT-5, and more
Built-in code editor, code review, file tree, and syntax highlighting… you won’t miss your IDE

See why Warp is trusted by 700k devs and 56% of Fortune 500 engineering teams.

👉 Try for free or get Pro for $1 with code NEURON1

Prompt Tip of the Day

Nano Banana is everywhere on social media as users pump out "professional grade" headshots. But we're liking more of Grace Leung’s top uses of Nano Bana for marketing.

The process is simple: upload your product image, then add a template design. Make sure the template goes last to preserve aspect ratio. You can generate 15 different product colorways in under 5 minutes before any physical production.

Three immediate applications come to mind:

Test makeup bag colors on social before manufacturing anything.
Generate lifestyle shots by specifying personas: “urban commuter with this backpack.”
Create 360-degree product videos using multiple generated angles.

Our favorite workflow: Find inspiring ad designs on Pinterest, feed them to a custom GPT that generates detailed scene prompts, then paste those prompts into Nano Banana. You get professional-grade product shots that match your vision instantly.

Treats to Try

*Asterisk = from our partners (only the first one!). Advertise in The Neuron here.

Cursor CEO shares: The smart way to scale when every hire delivers 10x | Michael Truell

*Rippling CEO Parker Conrad and Cursor CEO Michael Truell share their unfiltered playbook on scaling with AI. Discover why every hire's ROI will 10x and how to build sustainable moats when models are commoditizing. Live today only.
Cal.ai handles scheduling calls with human-like voice conversations and customizable scripts at $0.29 per minute.
Blocks lets you build automated AI agents that handle tasks like scheduling, data analysis, and emails across your existing apps.
Scaloom automates Reddit marketing by finding relevant subreddits, scheduling posts, and handling auto-replies to drive traffic.
Slashit creates reusable text templates with smart placeholders and hotkey shortcuts to eliminate repetitive writing tasks.
GraphBit provides an enterprise language model framework built with Rust for speed, security, and scalable AI agent development.
Check out these 9 “most sought after” AI startups from YC Demo Days. Our favorites are Autumn (Stripe for AI), Dedalus Labs (make agents in minutes), Design Arena (for help finding which AI is the best at design), and Pingo AI (learn languages through actual chat).

Around the Horn

OpenAI released the first detailed study of ChatGPT usage, revealing 52% of users are women and 73% of conversations are personal rather than work-related.
Reuters found major AI chatbots easily helped create phishing scams targeting seniors, with 11% of elderly volunteers clicking on AI-generated fraudulent emails.
AWS Managing Director Michelle Vaz said entry-level workers need curiosity and AI literacy to succeed as artificial intelligence reshapes rather than eliminates entry-level tech roles.
Fiverr plans to layoff 30% of its workforce as a new survey stated 60% of the tech industry is planning to restructure in the next 18 months (with options including M&A).
Religious chatbots like Bible Chat hit 30M downloads as people seek spiritual guidance from AI, despite concerns about validation bias.

FROM OUR PARTNERS

💼 Want to build a 6-figure AI Consulting career?

The AI consulting market is about to grow by a factor of 8X – from $6.9 billion today to $54.7 billion in 2032. But how do you turn your AI enthusiasm into marketable skills, clear services and a serious business?

Our friends at Innovating with AI have trained 1,000+ AI consultants – and their exclusive consulting directory has driven Fortune 500 leads to graduates.

Enrollment in The AI Consultancy Project is opening soon – and you’ll only hear about it if you apply for access now.

Click here to request access to The AI Consultancy Project