Neocloud accelerator
Plan your facility around demand. Work with NYTE to pre-sell reserved tokens from a planned facility to help finance the build, then deliver inference when it goes live.
Buyers start with a reserved tokens request; delivery, token accounting and commercial terms are agreed together.
Pre-selling means capacity reservations for inference service. This is a request to discuss a reservation, not an order. Capacity futures and derivatives are planned offerings, not live products. This is an inference service, not a securities or cryptocurrency exchange.
Both facilities are planned with targeted launches; NYTE lists them as reservation requests, not live products. Pricing is discussed on the call.
Operator names are shared on the call.
A planned solar-powered, air-cooled data center.
20 kW · Planned on NVIDIA B200/B300-class GPUs.
Targeted launch: February 2027.
First offering: 1,000 lots.
Grade B · Professional (DeepSeek-class models)
Each lot is 1 billion tokens delivered over the facility’s first twelve months of operation, not per month.
Discuss a reservation →A planned underwater data center.
1 MW · Planned on NVIDIA B200/B300-class GPUs.
Targeted launch: August 2027.
First offering: 50,000 lots.
Grade B · Professional (DeepSeek-class models)
Each lot is 1 billion tokens delivered over the facility’s first twelve months of operation, not per month.
Discuss a reservation →1B tokens over 12 months ≈ 32 tok/s sustained; 1,000 lots ≈ 32,000 tok/s; 50,000 lots ≈ 1.6 million tok/s.
NYTE supplies Neocloud-in-a-box, the Apache-2.0 open-source neocloud and inference stack. The software is free to use and pre-alpha. Run it yourself, or discuss having it remotely managed by NYTE. Remote management is an offer to discuss, not a live product.
NYTE connects you with demand from buyers on the exchange, with metering and billing through TrustedRouter. Hardware, deployment responsibilities and operating terms are scoped together. Confidential serving requires independent hardware and software qualification, not merely installing the software.
Bring your power, permits, building and cooling plan. Together we scope the serving stack, capacity and delivery requirements.
Define 1B-token lots over the planned facility’s first twelve months of operation on NYTE and discuss reservations with buyers. Agree the inference service, delivery conditions and payment terms before committing.
Build out the facility and deploy the stack. Complete the required qualification before serving buyer workloads.
Deliver the reserved inference service. Meter usage and settle through TrustedRouter under the agreed terms.
Discuss your facility, the reservation structure and how you want to run the software with Joseph.