OpenAI targets a late-2026 IPO with its nonprofit holding 26% & signs a multi-year AWS UltraServers pact. Plus: Cursor 2.0 has launched.
OpenAI is preparing for an IPO as soon as late 2026 valued up to $1 trillion, following a restructuring that gave its nonprofit foundation a 26% stake and more control while reducing Microsoft ties to fund its massive AI ambitions.
Mercor, now valued at $10B after a $350M raise, connects 30,000 global professionals to train AI models for clients like OpenAI and Anthropic. Its pivot from HR tech to data-labeling fueled rapid growth, boosted by Meta’s investment in rival Scale AI.
OpenAI and AWS formed a $38B, multi-year deal giving OpenAI access to AWS UltraServers with hundreds of thousands of NVIDIA GPUs to scale ChatGPT and future AI models, expanding compute through 2027 for faster, more secure AI growth.
Big Tech is investing $400B in AI this year to overcome compute shortages and chase AGI, with Meta, Microsoft, Google, and Amazon all boosting data-center spending. While investors back Google and Amazon, Meta and Microsoft face skepticism over returns.
Lambda signed a multibillion-dollar deal with Microsoft to deploy tens of thousands of Nvidia GPUs, deepening their AI infrastructure partnership amid rising compute demand and recent multibillion-dollar cloud deals by OpenAI, Microsoft, and AWS.
Canva launched its own design model that generates editable, multi-layered visuals, added AI-driven marketing tools through Canva Grow, and made its Affinity suite free and integrated, enhancing professional design, data visualization, and collaboration features.
Composer is a fast, MoE-based agent model optimized for software engineering through reinforcement learning, achieving top coding performance and 4× faster generation by efficiently solving real-world coding tasks within large codebases.
Cursor 2.0 adds multi-agent support, the fast Composer model, browser and terminal GA releases, voice control, improved LSP performance, and enterprise tools like team commands and audit logs.
SWE-1.5 is a frontier-scale coding model combining top-tier performance and 13× faster speed through RL training, co-designed infrastructure, and optimized inference on Cerebras, now available in Windsurf.
GitHub launches Agent HQ, a unified ecosystem integrating AI agents from partners like OpenAI and Anthropic directly into GitHub and VS Code, with mission control, enterprise governance, and code quality tools for seamless, secure AI-powered development.
OpenAI’s gpt-oss-safeguard-120b and -20b are open-weight reasoning models for customizable, explainable safety classification, outperforming prior models and enabling developers to apply and refine their own policies.
Codex’s investigation found no major flaw but several smaller issues—hardware lag, compaction inefficiency, and a constrained sampling bug—now fixed or being patched.
It's simpler, more stable, and doesn't require Node.js. They recommend this as the default installation method for all Claude Code users going forward.
Credits give you more usage beyond what’s included in your plan, kicking in when you hit limits.
It is an agent that finds and fixes security bugs using GPT-5.
Users will also get an option to Remix prompts from the feed.
The world’s first AI patent research agent that makes IP intelligence accessible to everyone.
Claude models show early signs of introspection, sometimes recognizing or controlling their internal states through “concept injection.” Though unreliable, this awareness is strongest in Claude Opus 4 and 4.1, hinting at growing AI self-monitoring capabilities.
After six months, the author built an automated Claude Code setup using hooks, skills, and dev docs to maintain context, with PM2 and agents ensuring error-free, consistent output across a 300k-LOC refactor.
A curated weekly roundup of AI news, tools, and breakthroughs to stay ahead of the curve in a fast-moving industry.
Curated by devs, for devs. Helping 400+ developers stay ahead of the AI curve.