Musk: Optimus Robot Production Will Be Painfully Slow at First Because Everything Is Brand New
On July 1, Tesla CEO Elon Musk shared a photo of himself on the Optimus humanoid robot production line at the Fremont factory in California, sparking fresh chatter about the bot’s mass production progress.
Some users speculated that Tesla’s recent cutback on public Optimus demos might mean production is ahead of schedule. Musk shot back, “No, Optimus production will be extremely slow at first because everything is brand new. It’s not like building cars.”
Alipay’s AI Assistant “Abao” Officially Opens Public Beta
On July 2, Alipay announced that “Abao” is now in public beta. iOS and Android users can just search “Abao” or “Ant Abao” in the app store or the Alipay app to try it out. The AI-powered version of Alipay “Abao” was first released on June 16 for invite-only testing.
SoftBank Plans New US Cloud Service to Meet Soaring AI Compute Demand
On July 2, SoftBank Group and SoftBank Corp. announced they’re forming a new company tentatively called SB Neo Inc., which will run a neocloud business in the US.
SB Neo will be set up in the US in July 2026, with SoftBank Corp. owning 51% and parent SoftBank Group holding 49%. Riding on SoftBank Group’s ongoing 10-gigawatt-scale energy and AI infrastructure push, the company will launch its neocloud service by the end of fiscal 2027, offering large US enterprises massive compute for AI model training and inference. To keep up with the insane US demand for AI compute, SB Neo will expand capacity in stages and eventually plan to deploy 10 GW of AI infrastructure.
According to NUPIAO sources, running this neocloud in the US could easily triple or quadruple SoftBank Corp.’s annual operating profit to a range of 3 trillion to 4 trillion yen ($18.5 billion to $25 billion).
Anthropic Reportedly in Talks with Samsung on Custom AI Chips
On July 2, NUPIAO learned from three sources that Anthropic has kicked off early-stage work on its own AI chips and held talks with Samsung Electronics about a potential manufacturing partnership.
According to the sources, Anthropic is still mapping out the processor’s features, compute power, and how it will sit in servers or clusters. The company has reached out to several chip design firms but hasn’t entered detailed design yet.
Xiaoyu Zhizao Launches First Embodied AI Welding Robot, Starting at ¥169,800
On July 2, at the 2026 Global Digital Economy Conference, Beijing Xiaoyu Zhizao Technology Co., Ltd. unveiled its industrial embodied AI welding robot, “Xiaoyu Future Robot.” The standard version starts at ¥169,800 (about $23,300) and packs three core technologies: the Xiaoyu Runwu World Model, Xiaoyu Xirang hardware platform, and a positioning pen. It’s already in use at over a dozen top-tier factories, covering steel construction, shipbuilding, and more.
Lenovo Exec: Long-Term AI Compute Demand Is Still Huge, No Oversupply in Sight
On July 2, NUPIAO reported that Lenovo Group VP and Chief Strategy Officer for Lenovo China, Ablikem Ablimit, said that in the long run, both in China and overseas, AI compute demand still has huge room to grow and there’s no oversupply. He pointed out that mature, top-tier AI products for consumers are still few, and the potential for B2B commercialization hasn’t been fully unleashed. Every past productivity revolution’s growth was driven by the industrial side, so long-term compute demand is massive. At the same time, he noted that the AI supply chain currently has a problem where the supply side’s voice is too strong, leading to an imbalance. This comes amid recent market jitters about compute oversupply triggered by Meta releasing its compute capacity.
NVIDIA Unveils New AI Infrastructure Partnership Model, Sharing Revenue with AI Cloud Providers
On July 1 local time, NVIDIA announced a brand-new AI infrastructure partnership model. It will use revenue-sharing and credit-support mechanisms to build large-scale, multi-tenant AI factories together with AI cloud providers. This aims to help startups, model developers, enterprises, and research institutions get AI compute faster, while also creating an ongoing revenue stream for NVIDIA tied to compute usage.
