Newton 1.0 goes fully open source

This is less a hobby repo than an attempt to standardize the simulation layer for robot learning. Linux Foundation governance matters here because robotics teams want a neutral stack they can trust, while the GPU-first design goes after the real bottleneck: running lots of high-fidelity sims fast.

• –Linux Foundation stewardship lowers vendor-lock-in concerns and makes the project easier to adopt across labs, startups, and hardware vendors
• –The Warp and OpenUSD foundation should make Newton fit naturally into NVIDIA Isaac workflows while still staying broadly usable
• –GPU parallelism is the point: Newton is optimized for large batches of simulations, which is what RL, domain randomization, and sim-to-real pipelines need
• –Support for multiple solvers, including MuJoCo Warp and Kamino, suggests this is aiming at more than a single narrow physics niche
• –The caveat is obvious: this is robotics infrastructure, not a general-purpose consumer launch, and it still assumes serious GPU-heavy workflows