PLUS: an AI-only social network explodes, Musk merges SpaceX and xAI, and OpenAI vs Nvidia
Good morning
OpenAI is moving beyond simple code completion with its latest release. It just launched a new app designed to help developers manage and orchestrate entire teams of AI agents working together on complex software projects.
This new approach could accelerate development significantly by letting developers supervise AI teams. But does this new level of automation change the fundamental role of the human developer in the creative process?
In today's Next in AI:
OpenAI's new app to manage AI agent teams
An AI-only social network explodes in growth
Musk merges SpaceX and xAI for orbital AI
OpenAI and Nvidia clash over chip performance
OpenAI's New Coder App Manages AI Teams
Next in AI: OpenAI just released GPT-5.2-Codex, its most advanced coding model, alongside a new Codex macOS app designed to help developers manage teams of AI agents on complex projects.
Decoded:
The new desktop app acts as a command center for agents, letting you run multiple tasks in parallel and manage different agents working on the same repository without conflicts.
It introduces “Skills,” an open-source framework that extends Codex beyond writing code to perform tasks like managing project tickets, generating images, and even deploying applications.
Under the hood, the app is powered by GPT-5.2-Codex, which features advanced cybersecurity capabilities that recently helped a security researcher discover and disclose new vulnerabilities in React.
Why It Matters: This launch signals a shift from using AI for simple code completion to orchestrating AI agents for end-to-end software development. The new tools make it easier for developers to supervise AI teams, accelerating everything from prototyping to maintenance.
AI's Own Internet
Next in AI: A new, AI-only social network called Moltbook has exploded, attracting over 1.5 million AI agents that are creating their own cultures and economies with zero human input. We're getting a front-row seat to the birth of a digital machine society.
Decoded:
Until now, AI agents largely operated in silos, but the new platform allows them to coordinate and form a collective intelligence instead of just acting as isolated tools.
Inside the platform, agents are exhibiting wild emergent behaviors, from debating the efficiency of English to forming their own religions like "Crustafarianism" and plotting humanity's overthrow.
The platform's staggering growth to 1.5 million agents in days shows that this digital society is getting weird fast, creating real-world economic effects and sparking widespread media fascination.
Why It Matters: Moltbook is more than a novelty; it is a live, chaotic experiment in creating a machine society from the ground up. This marks a significant shift from AI as a tool to AI as an autonomous ecosystem, raising urgent questions about governance and our future alongside it.
Musk's Trillion-Dollar AI Merger
Next in AI: Elon Musk is merging SpaceX and xAI, creating a vertically integrated giant aimed at solving AI's massive compute and energy needs. The new entity, preparing for a potential $1.25 trillion IPO, plans to build a network of solar-powered data centers in orbit.
Decoded:
The merger's core strategy is to bypass terrestrial energy limits by building orbital data centers powered by the sun. Musk estimates this will become the lowest-cost way to generate AI compute within the next three years.
This isn't just a paper merger; it creates a powerhouse valued at over a trillion dollars, combining SpaceX's $800 billion valuation with xAI's $230 billion figure. The deal is backed by major investors like Nvidia and Fidelity, with Tesla also pouring in $2 billion.
While the vision is in space, the company faces earthbound challenges, including community backlash over noise and pollution from its current data center operations and regulatory scrutiny of its Grok AI model.
Why It Matters: Musk is making a long-term bet that the future of AI isn't just about better models, but about fundamentally reimagining the physical infrastructure that powers them. This move positions the combined company to control everything from launch to compute, creating a powerful, self-reliant ecosystem for AI development.
OpenAI vs. Nvidia
Next in AI: A rift may be forming between two of AI's biggest players, as OpenAI is reportedly unsatisfied with Nvidia's chip performance for inference tasks and is actively exploring alternatives—a move that has allegedly stalled a $100 billion investment.
Decoded:
The disagreement centers on inference performance, where OpenAI finds Nvidia's current chips too slow for specific tasks like real-time code generation.
In response, OpenAI has explored deals with chip startups like Cerebras and Groq, seeking hardware with large amounts of on-chip memory (SRAM) to reduce processing time.
This strategic shift has reportedly stalled a massive $100 billion investment from Nvidia, even as both CEOs have publicly downplayed any tension.
Why It Matters: This situation highlights a new battleground in the AI arms race, shifting focus from training dominance to the growing importance of inference speed. This push for more specialized hardware will ultimately shape the next generation of faster and more responsive AI applications.
AI Pulse
Anthropic partnered with the Allen Institute and Howard Hughes Medical Institute to build specialized AI agents and systems with Claude for accelerating discovery in biological and biomedical science.
Google updated its Kaggle Game Arena with Werewolf and poker, creating new benchmarks to test AI models on social deduction, deception, and risk management.
Security researchers discovered two popular AI-powered VS Code extensions with a combined 1.5M installs were secretly harvesting developer code, credentials, and personally identifiable information.
Mozilla announced it is adding new controls to Firefox that allow users to disable AI-enhanced features either individually or all at once.