K-Scale Labs

What do they actually do

K-Scale Labs built an open-source humanoid robot called K-Bot and companion software for simulation, teleoperation, and training. They published hardware CAD, bills of materials, control stacks, and sim-to-real tooling on GitHub and in public docs so others could build, modify, and run the robot; early launch materials showed a Founder’s Edition and price messaging around $8,999 for initial units (kbot repo, docs, Launch HN thread, RS Online coverage).

In November 2025 the company shut down active operations after failing to secure follow-on funding, canceled and refunded pre-orders, laid off most staff, and open-sourced its remaining IP. The code and designs remain publicly available for independent builders and researchers, but the company is no longer shipping commercial units or providing vendor-backed support (Humanoids Daily, The Information, GitHub org, docs).

Who are their target customer(s)

• Academic robotics researchers needing an affordable, modifiable humanoid platform for experiments.: They need reproducible hardware and open tooling; with production halted, they must self-source parts and rely on community support rather than a steady commercial supplier (kbot repo, ksim-gym, shutdown coverage).
• ML/control engineers building sim-to-real locomotion or manipulation policies.: They need hardware that matches their simulators; K-Scale’s sim pipeline helps, but limited hardware availability and the shutdown made real-world validation and scale-up unreliable (ksim-gym, docs, The Information).
• Independent makers and hobbyists who want to build and learn from a humanoid design.: They face long lead times and complex sourcing; open CAD/BOMs lower barriers, but canceled orders mean builders must self-source and troubleshoot most of the supply chain (kbot repo, docs, Launch HN).
• Small robotics teams or startups wanting a low-cost off‑the‑shelf platform to prototype products or integrations.: They need predictable supply, pricing, and support; K-Scale aimed for this but hit funding/manufacturing roadblocks, creating acquisition and reliability risks for dependents (YC profile, The Information).
• Educators and technical training programs seeking a reproducible, documented robot for courses and labs.: They need safe, budget-friendly, well-documented platforms; public docs help curriculum-building, but lack of commercial units and support complicates class planning and maintenance (docs, GitHub org).

How would they acquire their first 10, 50, and 100 customers

• First 10: Directly recruit existing community members (repo star/fork lists, Discord, prior pre-orderers) with personal outreach, reserve Founder’s Edition units, and offer 1:1 assembly/onboarding using the public docs and sim stack (kbot repo, docs, Launch HN).
• First 50: Run funded pilots and conference workshops (ICRA/RSS/NeurIPS) targeting academic labs; sell a small number of assembled units or validated kits with a clear BOM and supplier links to ease replication, leveraging published examples for quick trials (ksim-gym, docs).
• First 100: Shift to small-batch manufacturing backed by deposits, partner with distributors for predictable supply, and rely on self-serve docs plus paid remote support—recognizing this depends on securing capital and production capacity, which K-Scale did not achieve before shutting down (docs, Humanoids Daily, The Information).

What is the rough total addressable market

Top-down context:

Long-run projections suggest a large potential market for humanoids (Goldman Sachs estimates ~$38B by 2035), and the broader robotics market is already sizable (ABI Research cites ~US$50B in 2025) (Goldman Sachs, ABI Research).

Bottom-up calculation:

For K-Scale’s near-term buyer set (labs, ML teams, educators), a bottom-up anchor is ~1,600 robotics-active universities; at ~$8,999 per unit, one robot each is roughly $14.4M (1,602 × $8,999) with upside if some buy 2–3 units, still modest versus the industry ceiling (EduRank, Launch HN pricing).

Assumptions:

• One unit per robotics-active university (1,600+ per EduRank) with limited industry lab uptake.
• Price per developer unit ≈ $8,999 based on launch messaging.
• Excludes supply/support constraints from the shutdown, which further reduce serviceable demand.

Who are some of their notable competitors

• Agility Robotics: Maker of Digit, a commercially produced biped with industrial deployments and a factory footprint—suited for teams needing turnkey hardware, spares, and vendor support rather than DIY builds (press).
• PAL Robotics: Sells complete research humanoid platforms (e.g., REEM-C, TALOS) with ROS integration and professional services for labs needing tested systems over open, self-built designs (REEM‑C).
• SoftBank Robotics (NAO): Widely used small humanoid for education and research with established purchase/support channels and curriculum materials; emphasizes reliability over hackability (reseller pricing example).
• Poppy Project / Poppy Humanoid: Open-source, 3D‑printable research/education humanoid with public CAD and an academic user community; philosophically closest to K-Scale but aimed at smaller, lower-power designs (GitHub).
• InMoov: Community DIY, life‑size 3D‑printed humanoid with STL files, BOMs, and build guides; fully community-driven with self-sourced parts and troubleshooting expectations (build/downloads).