Learn to train AI at home

​Training AI from your laptop? It's not the future... it's here.

Launching Training-at-Home

​Tuesday, 9th December, Macrocosmos AI Ltd is launching "Training-at-Home" through IOTA on the Bittensor network, enabling anyone with a consumer-grade MacBook to contribute to training frontier AI models and earn rewards for their participation. The breakthrough addresses a fundamental barrier that has prevented decentralised AI from competing with tech giants: the requirement for expensive, specialised hardware.

Come join us as we livestream the launch, meet the founders and developers and help shape the future of decentralised AI.

Livestreamed at: https://x.com/MacrocosmosAI
________________________________________________
ABOUT TRAINING-AT-HOME
Building leading AI models requires massive computational resources concentrated in data centers owned by a handful of corporations. Decentralized compute networks promised to break this monopoly but failed to solve the core problem because they still required users to own or rent expensive, high-end GPUs, simply creating a smaller, more complex version of the existing cloud market.

​IOTA's (a Bittensor subnet by Macrocosmos) technical advances in data parallelism, compression strategy, and model sharding lower the barrier to entry dramatically. Users need only a MacBook Pro or Air with 8GB memory and Apple Silicon chips, or Linux machines with 16GB VRAM and RAM, to participate in training runs for open-source models.

​"We're offering an alternative to the old model of centralized datacentres, which is like demanding that every company training AI builds their own power plant. Our breakthrough is more like plugging into a universal grid. For the first time, you don't need a data center; you just need the laptop already on your desk to help build and own the future of AI," said Dr. Steffen Cruz, CTO of Macrocosmos AI.