In partnership with
In Today’s Issue:
💻 Anthropic just launched full desktop autonomy
💼 OpenAI is aggressively courting private equity partners
⚡ Sam Altman is betting big on the future of AI infrastructure
🧠 GPT-5.4 Pro just cracked a notoriously difficult Hypergraph Ramsey math puzzle
✨ And more AI goodness…
Dear Readers,
Anthropic just handed Claude the keys to your Mac - and the entire software industry flinched. Yesterday's launch of full computer use in Cowork means your AI can now click, type, and navigate apps like a colleague sitting at your desk, and $285 billion in evaporated enterprise software value says this isn't a gimmick.
But that's just the opener: today we're also unpacking OpenAI's aggressive play to lock in private equity partners with a 17.5% minimum return, Sam Altman's audacious bid to secure 50 gigawatts of fusion power from Helion by 2035, and GPT-5.4 pro cracking a hypergraph Ramsey problem that stumped mathematicians for years. We've got fresh data showing Claude surging past DeepSeek, Grok, and Gemini to become the second most-used AI app daily, a roundup of robot milestones from Foxconn's factory floors to Unitree's IPO filing, and Roman Yampolskiy making the case that AGI will be humanity's last great invention.
Grab your coffee, this one's dense.
All the best,
🚀 OpenAI sweetens enterprise AI battle
OpenAI is reportedly dangling a standout 17.5% minimum return plus early access to its newest models to win private-equity partners, turning up the heat on Anthropic in a fast-moving fight for enterprise AI dominance. The bigger play is scale: these joint ventures could help both companies spread the heavy upfront cost of customizing AI across hundreds of portfolio companies, deepen customer lock-in, and build a stronger business story ahead of potential IPOs.
Not everyone is buying the pitch, though - some PE firms are questioning whether the economics, flexibility, and real upside are compelling enough, especially when many already have direct access to leading AI tools. That makes this more than a funding story: it is a high-stakes test of whether financial engineering can accelerate enterprise AI adoption at massive scale.
⚡ OpenAI Eyes Massive Fusion Power Deal
OpenAI is in advanced talks to secure up to 50 gigawatts of fusion energy from Helion Energy by 2035, starting with 12.5% of its output and scaling rapidly, highlighting just how urgent AI’s power needs have become. The deal, tied to Sam Altman-backed Helion, signals a bold bet on fusion as a clean, scalable solution, despite the company not yet reaching “scientific breakeven.”
If successful, this could redefine how AI infrastructure is powered - moving from energy constraints to abundance - but hinges on Helion achieving a historic technical milestone in time.
🧠 GPT-5.4 pro Cracks Hypergraph Ramsey Puzzle
GPT-5.4 pro has solved a tricky math problem that researchers had struggled with for years. The task involved finding better ways to build complex networks (called hypergraphs), and the AI discovered a more efficient method than previous approaches.
This is exciting because it shows AI can now help with real research problems, not just simple tasks, potentially saving experts months of work and accelerating new discoveries. This breakthrough can hardly be overstated: finding solutions that took human scientists years to achieve shows how quickly AI models are developing.
“Fast forward to today. I am so excited to release full computer use in Cowork and Dispatch. Really excited to see what you do with it!”
Anthropic keeps on shipping!
Roman Yampolskiy -THB Artificial General Intelligence Will be Humanity's Last Great Invention
Anthropic's AI Agent Now Controls Your Desktop, and That Changes Everything
The Takeaway
👉 Anthropic launched computer use for Claude Cowork on macOS, enabling the AI to click, type, and navigate apps autonomously, entering territory OpenClaw pioneered before OpenAI acquired its creator.
👉 Dispatch lets users assign tasks from their phone and return to finished results on their desktop, turning Claude into an always-available remote worker that operates across devices.
👉 The ripple effects are already massive: $285 billion wiped from enterprise software stocks since January, and Microsoft built its own Copilot Cowork product directly on Claude's technology.
👉 Currently limited to macOS Pro and Max subscribers in research preview, early adopters willing to experiment now will have a significant head start when this goes mainstream.
Anthropic just crossed a threshold nobody can ignore. As of yesterday, Claude can physically operate your Mac - clicking, typing, and navigating apps like a human sitting at your desk. The latest update to Claude Cowork and Claude Code makes the AI capable of opening files, using the browser, and running dev tools autonomously. Think of it as a remote colleague who never sleeps and never forgets your workflow.
The move puts Anthropic squarely in territory that OpenClaw, the open-source AI agent built by Austrian solo developer Peter Steinberger, pioneered before OpenAI snapped up its creator in February. Where OpenClaw gave power users raw, unguarded desktop autonomy, Anthropic is betting on a permission-first, connector-driven approach baked directly into its ecosystem. With Dispatch, the companion mobile feature, you can assign a task from your phone during your morning commute and return to finished work on your desktop.
The timing isn't accidental. Since Cowork's January research preview, enterprise software stocks have shed a combined $285 billion in value as investors realized what AI agents with deep tool integrations mean for traditional SaaS. Microsoft has already built its own Copilot Cowork feature directly on Anthropic's Claude technology, signaling that even Big Tech sees this as the new standard. Scheduled tasks, plugin marketplaces, Google Drive and Gmail connectors - Cowork is rapidly evolving from experiment to infrastructure. The real question isn't whether AI agents will reshape knowledge work. It's whether you'll be the one directing them - or watching from the sidelines.
Why it matters: Claude Cowork represents the clearest signal yet that AI is moving from conversation partner to autonomous digital worker. With OpenAI, Anthropic, Meta, and Microsoft all racing to own the agent layer, this shift will fundamentally redefine what productivity means for every knowledge worker.
Sources:
🔗 https://claude.com/product/cowork#dispatch-and-computer-use
Unlock AI-powered document editing with ONLYOFFICE Docs!
Boost your productivity with AI-powered document editing in ONLYOFFICE Docs. Connect your favorite AI assistant and summarize, analize, translate and do more right in your document.
Claude overtook DeepSeek, Grok, and Gemini to become the second most-used Gen AI app daily, after ChatGPT. Anthropic successfully bridges the gap between business tool and consumer software, and with the latest update, they've even proven that they can pick up on trends and develop their own (better?) products.
Overview of robot updates March 2026
(An employee works on a surface-mount technology (SMT) production line manufacturing electronics for key advanced driver assistance systems (ADAS) components)
Embodied Autonomy Hits the Road
Chinese autonomous-driving startup ZYT says its new “mobility foundation model” can already outperform its own CEO on Shenzhen’s chaotic streets, learning from video not just of roads but also drones, robots and even vacuum cleaners. The real breakthrough is the shift from brittle, map- and module-heavy autonomy to a more general embodied AI stack that could travel across cars, trucks and eventually other machines. That makes autonomy look far more scalable.
The Robot Brain Reaches the Factory Floor
Skild AI’s software is moving from lab promise into production: Foxconn will use it on Houston assembly lines building Nvidia Blackwell server racks, while ABB and Universal Robots will embed the same general-purpose model across industrial machines. What’s new is not another task-specific robot, but a reusable “robot brain” meant to transfer skills across hardware. If it works, factory automation gets cheaper, faster to deploy, and dramatically easier to scale.
Humanoid Robots Become a Public-Market Bet
Unitree’s planned Shanghai IPO is a signal that humanoids are becoming a capital-markets story, not just a demo-stage curiosity. Reuters reports the company shipped more than 5,500 humanoids last year, taking 32.4% of the global market, while humanoids became its main growth engine. The big idea here is industrialization: once investors start funding robot makers like mature hardware businesses, deployment could move from spectacles and receptions into real economic infrastructure.
Dictate code. Ship faster.
Wispr Flow understands code syntax, technical terms, and developer jargon. Say async/await, useEffect, or try/catch and get exactly what you said. No hallucinated syntax. No broken logic.
Flow works system-wide in Cursor, VS Code, Windsurf, and every IDE. Dictate code comments, write documentation, create PRs, and give coding agents detailed context- all by talking instead of typing.
89% of messages sent with zero edits. 4x faster than typing. Millions of developers use Flow worldwide, including teams at OpenAI, Vercel, and Clay.
Available on Mac, Windows, iPhone, and now Android - free and unlimited on Android during launch.