xAI doubles its round to $20B with $2B from Nvidia as it burns about $1B a month, Meta poaches Andrew Tulloch with a $1.5B package, and OpenAI expands AMD/Nvidia partnerships toward hyperscale. Plus: a deep dive on small‑sample data poisoning.
Sora surpassed ChatGPT’s iOS launch with 627K first-week downloads and hit 1M in under five days, despite being invite-only. It quickly topped the U.S. App Store, outpacing Claude, Copilot, and rivaling Grok in adoption.
Google’s Veo 3.1 enhances text-to-video generation with sharper visuals, better audio, and improved prompt accuracy, showing clear gains over Veo 3. Though not officially released, it’s poised to rival Sora 2 as testing ramps up via Google Vids.
xAI’s Grok is set to roll out tools for pulling and summarizing data from Gmail, Slack, Notion, and X, including financial info, aiming to streamline multi-platform insights. These upgrades push Grok toward a unified AI workspace experience.
xAI has named ex-Morgan Stanley banker Anthony Armstrong as CFO of both xAI and X, replacing Mahmoud Reza Banki amid a wave of executive exits, including its prior finance chief, general counsel, and co-founder Igor Babuschkin.
Andrew Tulloch, co-founder of Thinking Machines Lab, has joined Meta after being offered a $1.5B package, as Meta intensifies efforts to hire top AI talent following Llama 4’s disappointing results.
Elon Musk’s xAI is raising $20B—twice earlier estimates—to fund its Colossus 2 project, with Nvidia investing $2B and others backing a unique chip-rental structure, as the startup burns $1B monthly amid surging AI infrastructure demands.
OpenAI secured major infrastructure deals with AMD and Nvidia—worth up to $1T globally—as it builds toward self-hosted hyperscale AI, with Altman confirming more partnerships ahead despite revenue trailing its ambitious spending.
Cursor’s new Plan mode boosts coding with research-driven planning, inline editing, and Markdown file output, improving code generation and aiding complex tasks.
Google is releasing the Gemini 2.5 Computer Use model via the Gemini API, enabling developers to build agents that can interact with user interfaces. This model outperforms others in web and mobile control benchmarks with lower latency.
ChatGPT apps can now be built and deployed on Vercel using Next.js and MCP, enabling native integration with ChatGPT and leveraging Vercel’s full deployment stack.
Google Cloud hit a $50B revenue run rate, fuelled by AI growth. Gemini Enterprise, its new AI platform, integrates company data, tools, and agents to transform workflows, with real-world success seen at HCA Healthcare and Best Buy.
Amazon Quick Suite is a secure, enterprise AI assistant that connects to company data and tools to automate workflows, answer questions, and conduct research. Used by Amazon and clients like Jabil and DXC, it’s transforming productivity across roles and departments.
Users will be able to connect Claude app to GitHub and run their coding prompts on the go.
Plugins allow you to install and share curated collections of slash commands, agents, MCP servers, and hooks directly within Claude Code.
This model updates continuously to match ChatGPT, so it’s not always tuned for API use, but it works well for chat use cases.
OpenAI’s GPT-5 Pro now holds the highest verified frontier LLM score on ARC-AGI’s Semi-Private benchmark.
The new visuals are powered by Gemini’s image generation model, Nano Banana.
Duolingo tops the list, followed by OpenRouter, Indeed, Salesforce, and CodeRabbit.
A study found that injecting just 250 malicious documents can backdoor large language models up to 13B parameters, regardless of training data size. This challenges previous assumptions about data poisoning scale and highlights a serious AI security risk.
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.