KIDZ AI is pairing long-term contracted enterprise demand with purpose-built GPU infrastructure, a model it believes can serve as a repeatable template for the neocloud market.
Back
KIDZ AI signs $44.6 million GPU deal with Canopy Wave for neocloud push
KIDZ AI Inc. secured a $44.6 million, five-year GPU compute services agreement with Canopy Wave Inc., deploying 256 NVIDIA Blackwell B300 GPUs to power cost-efficient inference on open-weight AI models.
"This milestone contract is an important step in KIDZ AI's evolution into a broader AI technology and infrastructure platform," Stephanie Luo, chief executive officer of KIDZ AI, said.
Under the deal, KIDZ AI's subsidiary Catalyst Compute LLC will deploy 256 NVIDIA HGX B300 GPUs across 32 nodes, each with dual Intel Xeon 6776P processors, 4TB of DDR5 memory and 800Gb/s InfiniBand connectivity. Canopy Wave, an AI inference platform based in Santa Clara, California, will pay KIDZ AI $44.6 million over the initial five-year term for the compute services.
The agreement positions KIDZ AI, formerly known as Classover Holdings, in the neocloud segment alongside players like QumulusAI, which last week purchased 1,632 B300 GPUs in a single capacity expansion. KIDZ AI trades on Nasdaq under KIDZ and KIDZW, and the deal provides a contracted revenue stream tied to high-throughput inference on open-weight models — a segment the company expects to grow as cost per token becomes a key competitive differentiator.
A Differentiated Neocloud Model
Rather than deploying merchant GPU capacity and competing on raw scale, KIDZ AI pairs each infrastructure build with long-term contracted enterprise demand. The company purpose-builds for inference workloads on open-weight frontier models such as Moonshot AI's Kimi K3, a 2.8-trillion-parameter architecture, and DeepSeek. Canopy Wave specializes in optimized inference for these models, claiming industry-leading cost efficiency.
The approach mirrors a broader shift in AI infrastructure. Open-weight models have narrowed the capability gap with proprietary systems at substantially lower token cost, and KIDZ AI expects economic efficiency — cost per token, throughput, latency and utilization — to become an increasingly important differentiator as affordable AI expands the addressable market for compute.
Tao Zhang, chief executive officer of Canopy Wave, said the partnership is "just the beginning" and that his company aims to help KIDZ AI scale additional GPU clusters and turn the infrastructure into "token factories for the enterprise AI market."
What It Means for Investors
The $44.6 million contract gives KIDZ AI a five-year revenue pipeline anchored to real enterprise demand rather than speculative capacity. For context, QumulusAI — a publicly traded neocloud competitor — announced more than $124 million in three-year inference agreements earlier this year and has expanded its deployed GPU fleet by more than 450 percent from June 2025 to June 2026. KIDZ AI's smaller-scale deal signals it is targeting a focused niche: inference on open-weight models, where Canopy Wave's optimized platform provides a differentiated go-to-market partner.
Canopy Wave maintains SOC 2-certified data security and privacy controls, which Luo said positions the combined offering for enterprise customers seeking secure, cost-efficient AI inference with strong data protections. The agreement is subject to KIDZ AI's subsidiary placing a non-cancellable order for the GPU servers.
This article is for informational purposes only and does not constitute investment advice.
