Plus: Anthropic's first profitable quarter, OpenAI's IPO filing, and the $700M startup nobody saw coming
Sponsored by
Write, reply, and edit anywhere using Wispr Flow, an AI voice dictation tool that turns speech into polished text in any app, helping you write up to 4x faster with smart auto-edits and seamless cross-device support.
Try Wispr Flow FreeWelcome back to the world's #1 AI newsletter.
Today, we've got Google's massive I/O 2026 keynote dropping Gemini 3.5 and a wild new anything-to-anything model, Anthropic crossing into profitability with jaw-dropping revenue numbers, OpenAI racing toward an IPO, and the open-source AI agent project that turned down $20M to stay independent.
On this day in 2025…
Signs of AI model collapse were beginning to reveal themselves, as researchers sounded alarms about what happens when AI models are trained on AI-generated content. The Register reported on early evidence that recursive training on synthetic data could degrade model quality over time — a concern that has only grown more relevant as the industry scales.
🚀 Google I/O 2026: Gemini 3.5, Omni, and the Agent Era Arrive
Google dropped 13 major announcements at I/O 2026, headlined by Gemini 3.5 Flash (now the default model across Google products), the new Gemini Omni family that generates video from any combination of text, images, audio, and video, and Gemini Spark — a 24/7 AI agent that runs in the cloud across your entire Google ecosystem. With 900 million active Gemini users, Google is making its boldest play yet for AI dominance.
💰 Anthropic Hits First Profitable Quarter with $10.9B Revenue Surge
Anthropic's revenue is set to more than double to $10.9 billion in Q2 2026 — a 130% surge that will deliver its first-ever operating profit. The numbers, disclosed to investors during an ongoing funding round, defy skeptics who said AI companies couldn't turn a profit any time soon. The round could push Anthropic's valuation above OpenAI's.
📋 OpenAI Races Toward September IPO Filing
OpenAI is working with Goldman Sachs and Morgan Stanley on a draft IPO prospectus it plans to file with regulators in the coming weeks — possibly targeting a public debut as early as September. After raising $122 billion earlier this year, the AI giant now faces a high-stakes race to go public alongside SpaceX, which could be one of the largest IPOs in history.
🔥 xAI Burned $6.4 Billion Last Year — and Plans to Spend More
SpaceX's IPO filing reveals Elon Musk's xAI lost $6.4 billion on just $3.2 billion in revenue in 2025. Capital spending is on track to exceed $30 billion this year as xAI pushes Grok toward "multiple trillions of parameters" and eyes orbital data centers by 2028. Only 117 million of 550 million combined MAUs actually use Grok AI features.
🦄 Hark Raises $700M Series A for Its Secretive 'Universal' AI Interface
Brett Adcock's new venture Hark pulled in a jaw-dropping $700 million Series A at a $6 billion valuation for a product it has barely revealed. The company plans to release its first multimodal models this summer, followed by dedicated AI hardware — aiming to build the definitive personal AI platform.
⚡ NanoClaw is a security-focused, containerized alternative to OpenClaw that runs AI agents in Docker sandboxes instead of directly on your machine. Endorsed by Andrej Karpathy and Singapore's Foreign Minister, the project's creators turned down a $20M acquisition offer and raised a $12M seed instead — with execs at Amazon, Google, and Meta already using it.
🧑💻 Nemotron-Labs Diffusion is NVIDIA's new research into diffusion-based language models that promise "speed-of-light" text generation. Published on Hugging Face this week, this approach rethinks how LLMs generate text — potentially offering massive speed improvements over traditional autoregressive methods.
Coming in Hot
Free AI Video Generator, Image Generator, Text to Video & Avatar Maker
Open-source AI agent that learns and improves over time
Voice AI assistant for your website
Customer feedback and support platform for SaaS teams
AI photo editor and background remover
AI visibility tracking for marketing teams
Build and deploy AI agents without code
Run one-person companies with AI agents
Open-source web and product analytics platform
AI CFO that models startup decisions in minutes
Submit your AI tool to the #1 AI tools marketplace.
What it does: Mosaic is an agentic video editing platform that transforms hours of editing work into seconds using AI-powered Agents on a visual canvas. It combines localization, short-form creation, and chat-based control in one interface.
Why It's Worth Your Time
Pricing:
Why it matters: Video editing has long been one of the most time-intensive creative tasks. Mosaic's canvas-based, agentic approach feels like a genuine paradigm shift — letting creators think in terms of intent rather than individual cuts and transitions.
What it does: Hermes Agent is an open-source autonomous AI agent from Nous Research that lives on your own server or infrastructure, remembers what it learns, builds reusable skills, and becomes more capable over time instead of resetting every session. Unlike a coding copilot or simple chatbot wrapper, it keeps long-term context and can work across messaging platforms, automation workflows, and tools.
Why It's Worth Your Time
Pricing
Why it matters: Most AI assistants still behave like stateless chat interfaces that forget context after every conversation. Hermes approaches the problem differently by treating memory and learning as built-in capabilities. Instead of starting over repeatedly, it accumulates knowledge and workflows over time, making it closer to a persistent digital teammate than a temporary chat session.
🐝 Amazon's Bee Wearable Is Equal Parts Brilliant and Creepy
Amazon's AI wearable records and transcribes your conversations throughout the day, providing automated summaries and reminders. TechCrunch found it both genuinely useful and deeply unsettling — especially since all that data lives in the cloud. A local-processing version was teased but no timeline given.
🔎 AI Search Startups Are Blowing Up
Exa Labs raised $250M at a $2.5B valuation, Parag Agrawal's Parallel Web Systems pulled in $100M at $2B, and a wave of startups are chasing AI-native search. With ChatGPT owning the interface layer and Google protecting its ad business, there's room for nimble players to carve out serious market share.
🎙️ AI Used to Reconstruct Dead Pilots' Voices from Cockpit Recordings
People used AI on spectrogram images of NTSB cockpit recordings to reconstruct the voices of deceased pilots, forcing the safety board to temporarily block public access to its docket system. The intersection of AI capabilities and ethical boundaries continues to get murkier.
🏎️ Ferrari Is Using IBM's AI to Create F1 Superfans
IBM and Scuderia Ferrari are deploying AI to transform the Formula 1 fan experience — from real-time race insights to personalized content delivery. It's a fascinating look at how AI is reshaping sports entertainment beyond just analytics.
