KIDZ AI Inc. has signed a Memorandum of Understanding with Texas-based Limestone Networks, Inc. to colocate AI computing infrastructure at Limestone's DFW3 data center in Dallas, Texas, marking a concrete step toward deploying the hardware underpinning the education technology company's previously announced USD 44.6 million GPU compute services agreement with Canopy Wave, Inc.

The MOU, announced on August 3, 2026, outlines plans for a dedicated NVIDIA Blackwell B300 GPU cluster purpose-built for high-throughput, cost-efficient inference on open-weight AI models. The cluster will be ordered and deployed by Catalyst Compute LLC, a wholly owned subsidiary of KIDZ AI.

Infrastructure Designed Around Contracted Demand

The planned DFW3 deployment is intended to supply the power, physical space, network connectivity, and operating environment required to fulfill KIDZ AI's 60-month, USD 44.6 million GPU compute services agreement with Canopy Wave.

That contract, previously announced by the company, forms the commercial foundation upon which this infrastructure buildout is now being pursued. KIDZ AI Chief Executive Officer Stephanie Luo framed the sequencing as deliberate.

"We are not pursuing capacity on a speculative basis," Luo said.

"We secured long-term contracted demand first, and we are now putting in place the infrastructure required to serve it. That sequencing demand first, capacity second is the core of our neocloud model."

The company describes itself as pursuing a demand-led, inference-focused neocloud strategy, and the Limestone MOU represents what KIDZ AI characterized as the first data center step in executing on that approach.

Two Configurations Under Evaluation at DFW3

Under the terms of the MOU, the parties are evaluating two potential physical configurations at the DFW3 facility.

The primary option involves a dedicated data hall, designated Data Hall 150, offering approximately 1.125 megawatts of capacity.

An alternative configuration using two separate data halls is also being considered, contingent on confirmation that capacity, distance, and connectivity between the halls satisfy KIDZ AI's technical requirements.

If the single dedicated hall configuration is selected, initial deployment is expected to begin at approximately 0.5 to 0.6 megawatts.

The deployment could then ramp toward the full 1.125 megawatt capacity over time as GPU servers are delivered, installed, tested, and commissioned. The company noted that the configuration would also have room to accommodate additional GPU infrastructure as future customer demand develops.

The final configuration will be determined following further technical review.

Areas of ongoing evaluation include data hall allocation, power delivery, network connectivity, installation requirements, and related commercial terms.

Focus on Inference Compute for Open-Weight Models

KIDZ AI's strategic emphasis on inference compute, rather than training workloads, is central to how the company has positioned the planned cluster.

The GPU infrastructure is being purpose-built to support inference workloads for open-weight AI models, a segment of the market the company believes is experiencing the fastest growth in enterprise demand. Luo described inference compute as equally central to the company's identity as its demand-first capital allocation model.

"We are purpose-building this cluster for high-throughput, cost-efficient inference on open-weight AI models, which we believe is where enterprise demand is growing fastest," she said, adding that the approach is intended to support disciplined capital allocation, greater visibility into utilization, and what she called a repeatable playbook for subsequent GPU clusters and enterprise compute agreements.

MOU Terms and Path to Definitive Agreement

The MOU is described as non-binding, with the exception of customary confidentiality provisions. The parties have expressed intent to negotiate a definitive colocation agreement following completion of final technical design and capacity selection.

KIDZ AI noted in its announcement that there is no assurance a definitive agreement will be finalized or that the parties will be able to advance the transactions described in the MOU.

The proposed deployment remains subject to final technical design, confirmation of costs, and the negotiation and execution of definitive agreements.

About the Parties

KIDZ AI Inc., listed on the Nasdaq exchange under the tickers KIDZ and KIDZW, describes itself as an education technology company advancing AI infrastructure and GPU compute initiatives alongside its core EdTech operations.

The company states it is building capacity to serve enterprise and AI-native customers.

Limestone Networks, Inc. was founded in 2007 and operates as a cloud and data center infrastructure provider.

The Texas-based company offers services including Bare Metal Cloud, Cloud Instances, Private Cloud, and Colocation.

Limestone's colocation offerings range from individual server deployments to full racks and private cages, with infrastructure services available across multiple data center regions.