About NANDmind
Hardware designed to run the largest open AI models from flash.
Our mission
Make frontier AI models affordable to run anywhere — not only where GPUs and HBM are available.
Mixture-of-experts models use a small share of their weights for each token. We keep all the weights in flash and read only what each token needs.
Where we are
Two US provisional patent applications filed. A trace-driven simulator to validate speed and flash wear. The early-access list is open; the first batch follows the prototype.
Contact
Email
Investors
Investor materials are shared under NDA. For investors
Partners
Flash and controller makers: licensing and co-development. For partners
