GoodVision AI is diversifying its U.S. infrastructure strategy by pursuing two distinct deployment models designed to capture different segments of the AI inference market. Through new collaboration agreements with Tokenomics Digital Tech Corp. (TDTC) and HOP Global Service Inc. (HOP Global), the Redwood City-based company is testing whether it can simultaneously manage large-scale, dedicated "AI Factories" and smaller, more flexible computing clusters. These initiatives, centered in Texas, aim to anchor the company's software capabilities, such as its Smart Routing Engine, within a physical data center footprint. By bifurcating its approach, GoodVision AI is attempting to address the varying power availability and capital requirements found across the evolving American digital infrastructure landscape.
TDTC Modular AI Factory and 15MW Expansion Path
The collaboration with TDTC focuses on high-capacity, modular infrastructure designed for dedicated AI Factory operations. Under this agreement, GoodVision AI intends to provide the technical core—including GPU cluster design, network architecture, cooling systems, and compute-scheduling software—while TDTC manages site-specific requirements like power-resource development, permitting, and construction coordination. The initial Texas project, designated "Site No.1 AI Factory," is launching with 2.4 MW of confirmed power capacity, with an expected IT load of approximately 2 MW.
The architecture relies on prefabricated and containerized modules for power distribution, UPS, battery systems, and computing. This modularity is intended to create a replicable model for future power-ready sites. GoodVision AI expects to scale this specific project from its 2.4 MW starting point to a total capacity of 15 MW within one year. According to the company's internal financial models, achieving this 15 MW capacity could represent an annual revenue potential of approximately US$210 million. This high-density approach targets large-scale, dedicated compute requirements, providing a robust infrastructure anchor for both GoodVision and third-party customers once the facility becomes operational.
HOP Global RTX PRO 6000 Inference Pilot
In contrast to the heavy infrastructure of the TDTC deal, the agreement with HOP Global Service Inc. targets a "lighter" deployment model aimed at medium-sized Bitcoin mining enterprises. This pilot project utilizes RTX PRO 6000 hardware and 400G high-speed networking to validate smaller-scale AI computing clusters. The strategy is designed to provide a faster, lower-capital entry point for enterprises that possess existing power and sites but wish to avoid the immediate costs of large-scale data center construction.
The pilot will focus on conducting model inference workloads and multi-node performance validation to establish standardized delivery procedures. Each standardized computing node in this model requires approximately 0.05 MW of power, a relatively low threshold intended to support deployment in environments with limited power availability. GoodVision AI is currently evaluating the potential to scale this approach to 100 deployable nodes, which would represent roughly 5 MW of total computing capacity. If the company reaches this 100-node scale, its internal assumptions suggest an annual revenue potential of approximately US$36 million. This secondary path allows the company to test how AI inference infrastructure can be integrated into existing or adaptable computing environments.
Key Takeaways
- The TDTC collaboration aims to scale a modular Texas-based AI Factory from 2.4 MW to 15 MW within one year, with a projected annual revenue potential of US$210 million.
- The HOP Global pilot utilizes RTX PRO 6000 hardware and 400G networking to test smaller, 0.05 MW-per-node clusters for medium-sized Bitcoin mining enterprises.
- GoodVision AI is positioning these two distinct paths to address diverse infrastructure conditions, ranging from large-scale dedicated factories to incremental, low-capital deployments.
TechInsyte's Take
In our view, GoodVision AI is executing a pragmatic "hedging" strategy to navigate the massive capital and power constraints currently defining the AI infrastructure market. Rather than betting exclusively on massive, centralized data centers, the company is testing a dual-track approach: the TDTC model captures high-margin, large-scale enterprise demand, while the HOP Global pilot targets the "brownfield" opportunity of repurposing existing crypto-mining power and sites. This second path is particularly clever; it acknowledges that many enterprises have the power but lack the specialized AI architecture. By validating a 0.05 MW-per-node deployment, GoodVision is essentially attempting to commoditize AI inference at the edge of existing digital infrastructure. If successful, this flexibility could allow them to scale much faster than competitors who are tethered to the long lead times of traditional hyperscale construction.
Questions & Answers
How does the TDTC collaboration differ from the HOP Global pilot in terms of scale and target user?
The TDTC collaboration is a high-capacity model focused on large-scale, dedicated "AI Factories" using containerized infrastructure, aiming for a 15 MW capacity. The HOP Global pilot is a "lighter," incremental approach using smaller clusters (0.05 MW per node) designed to help medium-sized Bitcoin mining enterprises transition into AI computing without massive initial capital outlays.
What specific technical components is GoodVision AI providing in these partnerships?
For the TDTC modular factory, GoodVision AI is expected to provide GPU cluster design, network architecture, cooling systems, and compute-scheduling software. For the HOP Global pilot, the focus is on validating deployment architecture, 400G high-speed networking, and multi-node operations for inference workloads.
What are the projected financial implications of these two deployment paths?
According to GoodVision AI's internal models, the TDTC project has an estimated annual revenue potential of US$210 million once it reaches 15 MW. The HOP Global approach, if scaled to 100 nodes (approximately 5 MW), has an estimated annual revenue potential of US$36 million.
What is the primary objective of the RTX PRO 6000 pilot with HOP Global?
The pilot aims to validate the technical and operational requirements of deploying AI inference infrastructure within existing or adaptable computing environments. It specifically seeks to establish replicable technical solutions and standardized delivery procedures for smaller-scale, modular AI inference workloads.
Source: GoodVision AI
