DEPLOYDatabase

Company

Midcentury

www.midcentury.xyz

Physical AI data and simulation company.

Track Midcentury
Status
Private

Funding

$15.0M

Overview

Physical AI data and simulation company. Exited stealth in September 2026 with a $15M seed round. Claims a large labeled egocentric dataset and Matrix agentic simulation for policy evaluation and RL. Team background cited from Stanford, OpenAI, DeepMind, and NVIDIA. Dataset scale claims are company-claimed.

Reality check

Seed round (company-claimed)

Sep 2026: $15M.

Egocentric data (company-claimed)

2M+ hours of unscripted first-person video across 50+ environments, with hand pose, depth and task annotations.

Upcoming products (company-claimed)

MC-Motion (3D motion reconstruction), MC-Shade (photoreal rendering) and MC-PhysBench (physics benchmark), releasing soon.

MC-EgoHands (company-claimed)

Oct 2026: Turns first-person video into robot-learning episodes, exported as LeRobot v3 for Panda/LIBERO and Yam arms.

MC-EgoHands accuracy (company-claimed)

Oct 2026: Hand-pose error of 3.7-5.2 mm PA-MPJPE across four benchmarks, on the company's own evaluations.

Verified record

Verified deployments
None on file
Active incidents
None on file

Key facts

Focus

Physical AI data and simulation infrastructure.

Akari renderer (company-stated, Oct 2026)

Learned renderer that turns physics engine scene data into photoreal camera frames for robot sim, about 0.46 s per frame on one RTX 5090. Offered only through Matrix.

Akari test basis (company-run, Oct 2026)

Fidelity was measured against Unreal Engine renders, not real camera footage. Effect on real robot policies not yet tested.

Data & sources

Web sources

8

8 sources backing this record.View all →

Safety record

No incidents on record for Midcentury.

Only active incidents are counted. Retracted incidents are excluded from this summary but remain reachable at their canonical URLs.

Full safety record: incidents, sourcing, and exposure data →