Z Squared Inc. plans to build 100 megawatts of AI-ready data center capacity over the next 18 months, targeting the increasingly expensive market for AI inference workloads currently dominated by public clouds. The move, announced Tuesday, positions the digital infrastructure company to capitalize on firms seeking alternatives to unpredictable, token-based pricing for running their models.
"The company announced the Phase 1 destination of its strategy which includes a goal of reaching 100 megawatts (MW) of AI-ready capacity available for customer contracts, across multiple U.S. sites over the next 18 months," Z Squared said in a statement. The investment amount for the project was not disclosed.
The 100 MW build-out is specifically designed for inference, the process of running trained AI models to generate answers, which has become a major operational cost. A single developer using an agentic AI system can consume one billion tokens in 24 hours, costing an estimated $3,400 on a public cloud, according to recent figures from Dell Technologies.
Z Squared (NASDAQ: ZSQR) is betting that a significant share of these AI workloads will shift to specialized, fixed-cost infrastructure to escape such high costs. This pivot places it in competition with hyperscalers like Amazon Web Services and Microsoft Azure, aiming to capture a slice of the rapidly expanding AI infrastructure market before it fully matures.
The Economics of AI Inference
The push for alternative AI infrastructure stems from the punishing economics of agentic workflows. Unlike simple chatbots, AI agents that can run autonomously and retry failed actions consume tokens at a rate that makes public cloud billing a significant financial risk for enterprises. Dell, for instance, estimates its on-premises AI systems can reduce that spending by as much as 87% over two years compared to pure cloud solutions.
Z Squared's strategy is a direct response to this market opening. By offering dedicated capacity for inference, the company provides a predictable operational expense model, which is attractive for companies scaling their AI applications. The focus on inference is critical, as it represents the bulk of AI computational demand once a model is trained and deployed.
A Crowded Infrastructure Field
Z Squared enters a competitive arena where major technology players are already establishing their positions. Dell has become a key distribution channel for on-premises AI, forging partnerships to deploy OpenAI's models, Google's Gemini 3 Flash, and Palantir's Foundry platform directly on customer-owned hardware.
This makes the market for AI-ready infrastructure a battleground contested by traditional hardware vendors like Hewlett Packard Enterprise and Supermicro, and the hyperscalers themselves. Amazon's AWS and Microsoft's Azure are extending their own hybrid cloud offerings to keep enterprise workloads within their ecosystems. Z Squared's success will depend on its ability to build out its 100 MW capacity quickly and secure customers who prioritize cost predictability over the integrated services of the large cloud platforms.
This article is for informational purposes only and does not constitute investment advice.