OpenAI’s Big Move: In Talks to Lease a Massive 10-GW Data Center in Ohio
On June 10, NUPIAO reports that insiders are telling us OpenAI is seriously considering leasing a 10-gigawatt facility in Ohio. If they lock in a 20-year lease, this becomes their biggest infrastructure bet to date. Oh, and Nvidia is already circling the table to potentially back the deal with some serious credit support.
SpaceX and xAI Hit with Class Action: Over 10,000 Residents Say Elon Musk’s Projects Are Driving Them Crazy
NUPIAO has the details on this one: residents in Mississippi recently took SpaceX and its newly merged partner, xAI, to court. Why? The constant, ear-splitting noise from a power plant feeding their data centers has pushed locals to their limit. Filed documents from June 9 show three residents stepping up to represent over 10,000 people in federal court. They’re alleging both companies failed to dial down the racket, causing serious mental stress and tanking property values, and they’re demanding real compensation.
Anthropic Drops the Mic with Next-Gen Claude Fable 5
On June 9 ET, Anthropic officially rolled out its latest flagship model, Claude Fable 5, alongside a sibling release called Claude Mythos 5. Both run on the exact same core architecture. Here’s the thing: Fable 5 is the very first Mythos-tier model open to the public, while Mythos5 keeps the full, uncut capabilities locked away for a select group of trusted partners. Right on schedule, Anthropic also dropped the price tags for both: Fable 5 and Mythos 5 come in at $10 per million input Tokens and $50 per million output Tokens.
Moore Threads Opens Source MusaCoder: A Dedicated Code LLM Built Specifically for GPU Kernel Generation
On June 10, Moore Threads made a bold move by releasing and open-sourcing MusaCoder, a specialized code LLM designed to generate low-level GPU operators. According to the team, this is the industry’s first open-source code model trained and validated end-to-end on domestic GPU hardware. The entire post-training pipeline was executed on the Kua-E supercomputing cluster built around the MTT S5000 chips.
Zhiyuan Launches Lingxi X2 EDU “Build-It-Yourself” Version, Targeting Education & Training Scenarios
Also rolling out on June 10, Zhiyuan’s new Lingxi X2 EDU edition is squarely aimed at academic research, engineering training, and hackathons. Standing at 130cm tall and weighing 35kg, this bot packs 29 degrees of freedom across its body and 7 in each arm. It can carry a 3kg payload at its end-effector, sprint at 1.8m/s, and runs on an RK3588 compute module. Best part? You get full freedom to bolt on sensors and dev boards, plus open interfaces for joint and motion control, letting you tweak the underlying algorithms however you see fit.
He Xiaopeng’s Internal Memo: Taking the Wheel Personally for a Final 200-Day Sprint Toward Robot Mass Production
On June 10, NUPIAO has it that He Xiaopeng sent out a direct memo to his team, confirming he’s stepping in to personally oversee the robotics division. He noted that the human robot project is currently at a critical inflection point—comparable to where XPeng Motors stood eight years ago right before dropping its first car, the G3. We’re basically staring down the barrel of mass production. His roadmap is clear: Q4 2026 marks the start of production, Q1 2027 sees them rolling into retail stores to assist shoppers, Q2 2027 pushes them into international markets, and by 2028, we’ll start seeing these bots working side-by-side in households.
SpaceX Sets Its Sights on Space-Based AI Infrastructure Demos Starting Next Year
On June 10, reports indicate that SpaceX executives are laying the groundwork to debut initial space-based AI infrastructure demonstrations by late 2027. During a recent investor briefing, the company outlined its roadmap for showcasing orbital computing capabilities starting next year. Both President Gwynne Shotwell and CFO Bret Johnsen were present to walk through the ambitious plans.
Supermicro Announces Massive $7B Equity Financing Package to Fuel AI Server Demand
Last June 9, Supermicro announced it’s lining up a series of equity and equity-linked financing deals totaling roughly $7 billion. The company stated this cash injection is strictly to fund component procurement for the high-performance AI server orders they’ve been piling up lately. The package breaks down into a $5 billion underwritten offering (split between roughly $1.25 billion in common stock and $3.75 billion in depositary shares) and up to $2 billion in spot-market common stock sales, which could kick off as early as Q3 2026.
