A New Impetus for On-Premise AI Infrastructure

Orbital Industries, a startup with roots in London and San Francisco, has announced the closure of a $50 million Series B funding round. The operation, led by Plural and with participation from prominent investors such as Nvidia’s NVentures, Radical Ventures, Compound, and Fly Ventures, marks a significant step in the development of hardware solutions for data centers specifically designed for artificial intelligence workloads. The company, previously known as Orbital Materials, focuses on two key areas: PFAS-free cooling fluids and modular high-density compute infrastructure.

This investment underscores the growing demand for advanced physical infrastructure capable of supporting the escalating computational needs of LLMs and other AI applications. For organizations evaluating on-premise deployments, solutions like those offered by Orbital Industries become crucial for optimizing performance, energy efficiency, and control over their assets.

Innovation in Cooling and High-Density Compute

Orbital Industries' focus on PFAS-free (per- and polyfluoroalkyl substances) cooling fluids addresses a dual need: improving thermal efficiency and tackling growing environmental and regulatory concerns. Cooling is a critical factor in modern data centers, especially those dedicated to AI, where GPUs and other accelerators generate significant amounts of heat. An efficient cooling system not only prevents throttling and ensures optimal performance but also contributes to reducing the overall TCO by lowering energy consumption.

Concurrently, the development of modular high-density compute infrastructure is essential for maximizing space utilization and computational power within data centers. This approach allows companies to scale their LLM Inference and training capabilities more flexibly and economically, rapidly adapting to changing needs without resorting to massive infrastructure expansions or entirely cloud-based deployments, which could entail data sovereignty constraints and high operational costs in the long term.

Implications for On-Premise Deployments

The evolution of data center hardware, such as that developed by Orbital Industries, is of particular interest to CTOs, DevOps leads, and infrastructure architects who prioritize on-premise or hybrid deployment strategies. The ability to manage complex AI workloads in controlled environments offers significant advantages in terms of data sovereignty, regulatory compliance, and security. For sectors like finance, healthcare, or public administration, where information protection is paramount, the option to keep data and models within their own infrastructure boundaries is often a non-negotiable requirement.

Furthermore, optimizing hardware for AI can directly impact TCO. While the initial investment in on-premise infrastructure may be higher, an efficient and modular design can lead to significant long-term operational cost savings, including energy and maintenance, compared to cloud subscription models with variable and potentially increasing costs.

The Future of AI Infrastructure

Orbital Industries' funding reflects a broader trend in the tech industry: the growing awareness that the efficiency and sustainability of physical infrastructure are as important as software innovation for the advancement of artificial intelligence. The integration of AI-driven design into the hardware itself promises to unlock new levels of performance and resilience for future data centers.

For companies facing the choice between on-premise deployments and cloud solutions for their AI workloads, the emergence of technologies like those from Orbital Industries strengthens the argument for greater control and customization. AI-RADAR, through its analyses and Frameworks available on /llm-onpremise, continues to explore the trade-offs and opportunities these innovations offer for informed strategic decisions in the AI landscape.