TL;DR
INDUSTRY WATCH #1
Nov 05, 2025
XPENG unveils new “IRON” humanoid robot
At XPENG’s AI Day the next-generation IRON humanoid robot was revealed, walking across the stage in a strikingly human-like way. The robot has 82 degrees of freedom (DOF), 22 DOF in each hand, three in-house AI chips delivering about 2,250 TOPS (trillions of operations per second) compute power, a human-style spine and bionic muscles.
The robot exhibited a catwalk-style demonstration with fluid human-like movement, prompting speculation that a human might have been inside the prototype. The company then posted another video showing its insides to prove it is robotic. The initial video went viral with a lot of questions around the need of the female form in humanoids.
This launch signifies that humanoid robots are entering the “show stage” in major tech firms outside of robotics, pushing expectations for physical AI beyond labs and into real-world human form. But the risk is that demonstration mobility may not scale into reliable, safe general‐purpose deployment.
INDUSTRY WATCH #2
Oct 21, 2025
Tesla shareholder meeting includes Optimus strategy and roadmap
At its annual meeting, Tesla previewed its humanoid robot line, Optimus, including production targets and strategic framing of robots as a major future product. Key claims include: a pilot line capable of ~1 million units per year, future facility in Texas targeting ~10 million units per year, and a long-term unit cost target of ~US$20,000 once volume production is reached.
Tesla framing humanoids as a “biggest product of all time” signals a shift from vehicle-only strategy to robotics. The risk: scaling humanoids to millions of units and achieving low cost is a massive industrial and engineering challenge.

Optimus production line (Image Credit: Tesla)

Optimus production line (Image Credit: Tesla)
IN FOCUS
Nov 04, 2025
NEURA Robotics secures industrial partnerships and funding
German startup NEURA Robotics announced two major items this week. First, a technology and offtake partnership with Schaeffler AG: NEURA will supply humanoid robots and Schaeffler will integrate a mid-four-digit number of units into its global production by 2035.
Secondly, NEURA partnered with SAP to pilot a warehouse humanoid system: a robot “4NE1” was trained in NEURA’s lab using SAP systems and Sartorius product handling.
Also last week, several business-news outlets reported that NEURA Robotics is planning a new funding round of up to ~€1 billion (US$1.2 billion) to support production of its humanoid robot platform.
NEURA is anchor-positioning itself at the industrial edge of humanoid deployment, partnering with major manufacturing firms and enterprise software platforms. It shifts the narrative from “future robot concept” to “robot in factory/logistics workflow.” That said, these are pilots, not mass production—so the path to scale remains open.

Neura Robotics working in BITZER warehouse operations
DEEP DIVE
K‑Scale Labs shuts down
K-Scale Labs, a Palo Alto-based startup developing a low-cost open-source humanoid robot called “K-Bot”, has shut down operations. According to the company’s CEO in an investor email, the firm was unable to secure further funding and will refund all customer pre-orders. The startup has also released its hardware and software IP under open-source licences.
K-Scale’s mission centered on “democratizing” humanoid robotics by offering an open-source stack and hardware for developers, researchers and hobbyists. Their flagship robot, the K-Bot, was priced at around US $8,999 and stood approximately 4′7″ tall, with the Z-Bot model targeting educational/hobbyist markets at around US $1,000. The company reported over 100 pre-orders and had developed ten prototypes, yet could not scale manufacturing or secure enough capital to move into volume deployment.
The shutdown highlights the immense capital and engineering challenges in scaling humanoid robotics. Even with a compelling open-source model and early traction, K-Scale could not bridge prototype to production. The ripple effect: investors and startups may become more cautious about humanoid ventures without clear paths to scale. The open sourcing of its IP may benefit hobbyists and academic labs, but it underscores that commercial humanoid deployment remains expensive and high risk.
Editor’s Take
This week marks a step change for humanoids. On one front we saw XPENG push the flashy demo envelope—“walks human” is no longer the hurdle. On another, Tesla laid out its mass-production vision for humanoids. Not just as research prototypes but as scaled products. And mid-tier we have NEURA quietly securing industrial commitments that reduce the “science project” stigma for robots and move towards real workflows.
If the industry pulls this off the next 18-24 months, we might be past “could humanoids work?” and into “which one will work first and where?”
Optimus is a fundamental part of sustainable abundance
