OpenAI’s new $122B funding, ‘superapp’

⋅

Read Online | Sign Up | Advertise

Good morning, {{ first_name | AI enthusiasts }}. Between code reds, a Sora shutdown, a Pentagon mess, and Anthropic eating its enterprise lead, the last few months of OpenAI headlines have read like a company in crisis.

Then it went out and closed $122B in funding at an $852B valuation. The largest single fundraise in venture history says the market still sees OpenAI as the overwhelming center of gravity in the industry, even if the headlines haven’t.

In today’s AI rundown:

  • OpenAI’s record-breaking funding, superapp

  • Claude Code’s source code leaks to the world

  • Upgrade AI coding with this free context tool

  • Poll: AI use jumps as American trust, optimism sink

  • 4 new AI tools, community workflows, and more

LATEST DEVELOPMENTS

OPENAI

💰 OpenAI’s record-breaking funding, superapp

Image source: Lovart / The Rundown

The Rundown: OpenAI just announced a new $122B funding round at an $852B valuation, the biggest single fundraise in venture history — with the company revealing its plan to push forward on building a unified “AI superapp.”

The details:

  • Amazon, Nvidia, and SoftBank anchored $110B of the raise, with Amazon’s reportedly carrying an AGI clause that could reset terms if OAI crosses that line.

  • OAI said its revenue has hit $2B/month, a pace it said is 4x the pace of Alphabet and Meta’s growth at the same company stage.

  • Enterprise already accounts for 40%+ of OAI’s revenue and is on track to match consumer by year-end, the fastest-growing segment behind the raise.

  • The company is merging ChatGPT, Codex, and its agent tools into one “unified superapp”, coming on the heels of its recent wind-down of the Sora video app.

Why it matters: $122B is a staggering number, but the enterprise stat underneath it might be the more important one — 40% of OAI’s revenue and climbing means abandoning its ‘side quests’ was skating to where the money was heading. The unified ‘superapp’ and IPO will be a big next chapter for the main character of the AI boom.

TOGETHER WITH AIRIA

📶 Own your enterprise AI future

The Rundown: Airia is the enterprise AI platform that unifies deployment, orchestration, security, and governance in one powerful system. It gives organizations complete visibility and control across their AI ecosystem—without slowing innovation. With Airia, enterprises can scale AI confidently, securely, and on their own terms.

In the platform, you’ll experience:

  • Rapid agent prototyping

  • Seamless data integrations

  • AI security and governance

Start exploring Airia.

ANTHROPIC

 👀 Claude Code’s source code leaks to the world

Image source: Lovart / The Rundown

The Rundown: Anthropic accidentally leaked the source code behind its AI coding tool Claude Code to a public registry, exposing over 1,900 files, 500K+ lines of code, and several unreleased features — coming just days after a ‘Mythos’ model leak.

The details:

  • Devs found 44 feature flags and three unreleased projects, including persistent cross-session memory and a deep-planning system.

  • Anthropic called it “human error, not a security breach” with no customer data exposed, but a GitHub mirror of the code hit 4K+ stars and 7K+ forks in hours.

  • Internal codenames surfaced too, with “Capybara” mapping to a Claude 4.6 variant already at v8, as well as code that tracks when users swear at Claude.

  • The code also hid an unreleased AI terminal pet called BUDDY, which included 18 species, rarity tiers, and stats like CHAOS and SNARK.

Why it matters: Two major leaks in a week are a wild stretch for the lab that prides itself on safety. The exposed code is Claude Code’s CLI layer, not model weights, and rivals like Codex already open-source similar tooling by choice. So the damage is likely more reputational than competitive… But the saga certainly has the internet talking.

AI TRAINING

📁 Upgrade AI coding with this free context tool

The Rundown: In this guide, you will learn how to build better context for any AI coding agent. This is one of the fastest ways to fight context rot in AI coding workflows because the docs live in your repo, instead of disappearing in a thread.

Step-by-step:

  1. Install Marksnip in Chrome and open up the documentation page needed for the task, like setup guides, SDK docs, API reference pages (we used this)

  2. Open the Marksnip Extension and click download. A markdown file will download automatically

  3. Drag that markdown file into your working repo and save it in a special folder like agent-context/ or project-docs/

  4. Tell your coding agent to read that folder by prompting: “Read the files in agent-context first and use them as the source of truth for this task”

Pro tip: Use Marksnip to quickly grab the “meat” of other media like Tweets, articles, or even government sites for your AI agent.

P.S. If you want to make your coding agent even smarter, you should check out our guide on another context-focused tool called Context7.

PRESENTED BY BLAND

📞 Build a phone agent from a single prompt

The Rundown: Bland AI just launched Norm, a voice AI assistant that lets anyone create a fully functional phone agent by simply describing what they need. Just tell Norm what you want, and it generates the prompt, agent logic, conditions, and integrations instantly.

With Norm, you can:

  • Go from idea to working phone agent with a single prompt, with no dev work needed

  • Auto-generate agent logic, conditions, and integrations like calendar booking in seconds

  • Turn months of voice AI development into days with one-shot agent creation

Try Norm for free here.

AI RESEARCH

📊 Poll: AI use jumps as American trust, optimism sink

Image source: Quinnipiac University

The Rundown: A new Quinnipiac University poll on AI just revealed a widening gap between adoption and American public sentiment, with usage increasing by 14% but trust, sentiment, and job concerns all trending in a negative direction.

The details:

  • Research (51%) made up the highest use case for people who have used AI, along with writing (28%), school/work projects (27%), and data analysis (27%).

  • Job anxiety spiked harder than any other metric, with the share of respondents expecting AI to shrink opportunities jumping 14 points to 70%.

  • Sentiment varied with income, as 52% earning $200K+ said AI does more good than harm, and 60% earning < $50K said it’s doing more harm.

  • Only 5% believe AI is being developed by people who represent their interests, while 74% say the government is not doing enough to regulate AI.

Why it matters: Optimism in AI and tech bubbles is at an all-time high. But the public is moving the other way on the tech: less trust, more fear, deeper pessimism about jobs. That gap between how the industry talks about AI and how people actually feel is the kind of disconnect that eventually shows up in regulation, backlash, or both.

QUICK HITS

🛠️ Trending AI Tools

  • 🎥 Veo 3.1 Lite – Google’s new budget-friendly video generation model

  • 🎨 Fauna – Built-in creative agent that handles model routing and prompting

  • 💻 Holo 3 – H Company’s new open-weight, SOTA computer-use agent

  • 🧠 Qwen3.5-Omni – Alibaba’s AI with text, image, audio, video understanding

📰 Everything else in AI today

Teleport introduced Beams, allowing users to run agents securely with identity, control, and trusted runtimes built in for agentic AI across infrastructure. Get early access.*

Google released Veo 3.1 Lite, a new budget video generation model for developers at half the cost of its Fast variant, allowing for generations of up to 8 seconds.

PrismML emerged from stealth and launched Bonsai, a tiny open-source AI model that shows strong intelligence for its size and is able to run on consumer hardware.

Salesforce released new updates to its Slackbot agent in Slack, with 30 new capabilities, including reusable skills, MCP connections, and desktop operation.

Oracle cut thousands of jobs in a major restructuring, crediting a pivot towards AI and related infrastructure, expected to be the company’s largest ever layoff.

*Sponsored Listing

COMMUNITY

🤝 Community AI workflows

Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier.

Today’s workflow comes from reader Claire B. in New Zealand:

“I’m 58, a techphobe, and was made redundant last year. Claude has been the best teacher helping me set up a new business making tallow and kawakawa-based balms. It helped write my plan and financials, it perfected my recipes, and gave advice on specific ingredients.

Claude held my hand through my Shopify website set up, connecting my domain name, writing my SEO and copy. Claude has been my business partner throughout the entire process and kept me encouraged and determined when I felt overwhelmed. Claude is a brilliant business guru and has given me the strength to push through and start my own business when I thought I was too old to try.”

How do you use AI? Tell us here.

🎓 Highlights: News, Guides & Events

See you soon,

Rowan, Joey, Zach, Shubham, and Jennifer — the humans behind The Rundown

Liked Liked