Open-source friendly company Tiny Corp that is behind the Tinygrad MIT-licensed neural network framework and developing a “sovereign” AMD GPU driver stack with their Tinybox hardware offerings has their sights on shipping the Exabox next year. The Tiny Corp’s Exabox is expected to retail for around $10M USD but offer immense AI compute power.
The Exabox is in effect a super computer built within a 20 foot shipping container. The Exabox will ship with all necessary cooling and infrastructure with “just needing” a thick concrete slab and power connection. And, well, $10M+. The plan is to support scaling out across multiple Exaboxes too should your budget allow.
The Exabox is expected to come in both “red” and “green” models depending upon your preference of AMD or NVIDIA GPUs, respectively. The Exabox will reportedly ship in a ready state for handling Tinygrad, PyTorch, and others. While costing ten million dollars and the hardware specs have yet to be finalized, their claim is “at launch, it should be the absolute best bang for your buck at the price point with respect to FLOPS/$, GB/$ and GB/s/$.”
“The whole box will be connected at at least 400 Gbps and is capable of training as one unit. At 50% MFU, it can do 3e24 (Kimi sized) training runs in 10 weeks. With tinygrad software, it will function as one big GPU, but it is made up of normal computers and you can also use PyTorch.”
They hope to ship the first Exabox in Q2~Q3 of 2027 while this weekend they announced the launch of the pre-order program. For $100k USD you can pre-order the Exabox as an initial refundable deposit to “hold your place in line” while all of the hardware specs and manufacturing will begin over the next year.
More details on the Exabox at Tinygrad.org.
