Sam Altman Hints OpenAI Could Go Public Within a Year
On June 8, OpenAI CEO Sam Altman dropped a bombshell in a Slack message to his team, saying the company is looking to go public “within the next year.” He noted that the exact timeline might shift a bit, but “filing now gives us more flexibility.” On the very same day, OpenAI confirmed they’ve secretly filed a draft S-1 registration for an IPO with the SEC.
OpenAI Reportedly Mulling Massive Price Cuts to Poach Anthropic’s Users
Fast forward to June 11, and we’re hearing whispers that OpenAI is thinking about slashing prices big time to steal customers from rival Anthropic. But here’s the catch: these steep discounts could seriously eat into both companies’ profit margins. And let’s be real, they’re already bleeding billions thanks to the massive computing costs needed to run these AI systems.
Xiaomi Drops Open-Source AI Coding Assistant MiMo Code
Also on June 11, the Xiaomi MiMo tech crew officially launched their AI coding buddy, MiMo Code. Built on the open-source OpenCode project, it packs some serious features like persistent memory, unlimited context, agent collaboration, and a unique Compose mode. It plays nice with major models like DeepSeek, Kimi, and GLM. Plus, it’s open-sourced under the MIT license and even throws in their multimodal MiMo V2.5 model for free—well, for a limited time, anyway.
Microsoft President Addresses Graduates’ Fears: AI Will Empower, Not Replace Us
On June 10, Microsoft’s Vice Chair and President Brad Smith shared some thoughtful insights, noting that while AI is going to shake up our economy and society, the tech world can’t just brush off young people’s worries about their job prospects. He pointed out that recent pushback from US college grads against AI hype really shows that folks want humans—not machines—calling the shots on how AI is used.
He made it clear that even though AI will take over some tasks and roles, it’s going to take a while for the tech to spread and truly impact the job market. His advice? Break your work down into three buckets: “what AI can do,” “what humans and AI can do together,” and “what only humans can do.” Then, double down on uniquely human traits like curiosity, creativity, empathy, communication, and good old-fashioned courage.
German Robotics Firm Neura Robotics Scores $1.4B Funding with Backing from Amazon and Nvidia
On June 10, German robotics outfit Neura Robotics announced a record-breaking Series C funding round, pulling in up to $1.4 billion to fast-track their mission of building a world-class physical AI platform. This massive cash injection is backed by heavy hitters like Amazon, Nvidia, and the European Investment Bank.
Guangxiang Tech Unveils Industrial Self-Evolving Embodied AI Robot Phi-Bot X1
On June 10, 2026, Guangxiang Tech officially rolled out the Phi-Bot X1, the industry’s first industrial-grade, self-evolving embodied AI robot. Built for high-precision, multi-station, and fast-paced industrial environments, the X1 features a robust industrial design and runs on their proprietary physics-native AI model. It also leverages the Phi-Arch physical intelligence dev platform for scalable deployment. With core strengths like “multi-tasking, quick onboarding, and self-learning,” this bot can autonomously perceive, make decisions, operate, and keep evolving right on the real production line.
OpenAI Teams Up with Oracle: OCI Customers Can Now Use Cloud Credits for OpenAI Models and Codex
OpenAI and Oracle announced a new partnership on June 10. This means Oracle Cloud Infrastructure (OCI) customers will soon be able to tap into OpenAI’s cutting-edge models and Codex using their existing Universal Credits Model (UCM) commitments. The best part? It’ll be available in just a few weeks, and you won’t need to jump through any extra procurement hoops.
Google Drops the DiffusionGemma Open-Source Model
On June 10, Google released an experimental open-source model called DiffusionGemma. It uses a text diffusion architecture and can generate text up to 4 times faster than traditional autoregressive LLMs on dedicated GPUs. It’s out now under the Apache 2.0 license. Google was pretty upfront, though: DiffusionGemma is strictly an experimental toy for researchers and devs. Its overall output quality doesn’t quite match the standard Gemma 4, so stick with Gemma 4 for production. The real speed boost shines in local or low-concurrency setups, but don’t expect miracles in high-concurrency cloud deployments.
Anthropic CEO Calls for Mandatory AI Testing and Government Intervention
Also on June 10, Anthropic CEO Dario Amodei penned a lengthy post arguing that governments should have the power to stop AI companies from deploying new models if they pose specific risks. Amodei is pushing for mandatory third-party testing to evaluate potential dangers across various fields, like sparking cybersecurity threats or even creating bioweapons. His bottom line? If an AI is deemed to pose an “unacceptable risk,” the government absolutely should have the authority to block or restrict its rollout.
