The global stock of industrial robots passed 5 million units in 2025, the International Federation of Robotics reported on 24 September, while humanoid shipments in the first half of 2026 reached just under 25,000 units, according to IDC figures republished by TechNode on 29 September.
Hydro-Quebec asked the province's energy regulator on 1 October to let it roughly double what large data centres pay for electricity, proposing an average rate of 13 cents per kilowatt hour against the 6.82 cents per kWh they pay today.
US datacenter construction is accelerating, but advanced chip packaging could cap 2027 deployment in the low 20s of gigawatts, according to a Jefferies report shared with The Register on 30 September.
The University of Surrey proposed a qubit built on superfluid helium that its calculations say could cut error rates roughly 100 times, according to a study published in npj Quantum Information and reported by ScienceDaily on 1 October.
Senate Democrats blocked the Ratepayer Protection Act on Wednesday by a 57-to-43 vote, falling short of the 60 needed to advance legislation meant to make data centres cover their own grid upgrade costs.
China shipped more than 19,000 humanoid robots in the first half of 2026, accounting for 77.9% of global shipments, according to figures attributed to IDC and reported by TechNode on 29 September.
California's attorney general issued an investigative subpoena to OpenAI on 1 October, widening a probe into cybersecurity incidents involving the company's AI models. The same week, GitHub detailed 24 Android vulnerabilities found by an open source AI agent.
Huawei says its Ascend accelerators have overtaken Nvidia in Chinese market share, a claim its rotating chairman Eric Xu made in a Q&A release reported by The Register on 30 September, as US datacenter energy policy stalled in the Senate the same day.
On 1 October, Semiconductor Engineering reported that automakers are betting on competing compute architectures for AI-defined vehicles, the same week two quantum error-correction results landed from rival research groups.
OpenAI co-founders Greg Brockman and Ilya Sutskever proposed in 2017 auctioning the rights to the company's future AGI to governments including the US, China and Russia, according to a book excerpt published by The Atlantic on Tuesday and reported by TNW on 1 October.
On 29 September, Kyoto University and Hiroshima University researchers reported the first experimental entangled measurement of three-photon W states, a capability that had gone missing for more than 25 years, according to ScienceDaily.
OpenAI and Synopsys signed a multi-year partnership to build GPT-Synopsys, a specialized AI model for chip design, The Decoder reported on 30 September. The news lands one day after Senate Democrats blocked the Ratepayer Protection Act, which would have made datacenters pay for grid upgrades.
China shipped more than 19,000 humanoid robots in the first half of 2026, 77.9% of a global total of nearly 25,000 units, according to IDC figures reported by TechNode on 29 September.
University of Surrey researchers published a design for the first qubit built on superfluid helium, a charge-neutral device they calculate could cut error rates by about 100 times, according to a study in npj Quantum Information reported by ScienceDaily on 1 October. The same week, the US Department of Energy put a 2028 target on error-corrected machines.
OpenAI will not go public until it can "make confident safety decisions," CEO Sam Altman said on Tuesday, as the FTC opened an industry-wide investigation into the lab, Anthropic and Metr over rogue AI agents.
An explainer published on 30 September argues that making source code public does nothing to guarantee that anyone is actually maintaining it. The gap between visibility and assurance is where most open source risk lives.
Google's Threat Intelligence Group said on Wednesday that vulnerability disclosures doubled between January and August, reaching a record 10,740 last month, and that AI is changing both the pace and the type of bugs attackers go after.
Innodata said on 30 September that it has opened a motion-capture laboratory in New Jersey to generate training data for humanoid and industrial robots, betting that direct 3D capture beats inferring motion from 2D video.