The rise of artificial intelligence and the rapid expansion of hyperscaler data centres are reshaping the Real Asset investment and operation landscape. In this context, Taronga Group has invested in the oversubscribed round for Salience Labs – a University of Oxford spin-out developing breakthrough optical switching technology for AI data centres.

Rapidly growing demands for data centre compute and energy

The rapid global emergence of AI has accelerated the build out of ‘AI data centres’ which have meaningfully differentiated compute, energy and infrastructure requirements. According to McKinsey, global data centres will require approximately US$6.7 trillion in capital investment by 2030 to mainly meet demands for AI-workloads (McKinsey, April 2025).

This unprecedented growth is applying significant pressure on resources. Computationally-intensive AI workloads – up 10x more dense racks than traditional data centres – are growing at 33% per year and estimated to represent 70% of global data centre compute by 2030 (McKinsey, October 2024). From an energy perspective, US data centres consumed 176 TWh (4.4% of total) in 2023, with that figure projected to rise to between 325–580 TWh (6.7–12% of total) by 2028 (2024 United States Data Center Energy Usage Report, Lawrence Berkeley National Laboratory, 2024). Similar energy constraints are forecast in key data centre markets globally.

These factors are driving data centre developers to identify sites with sufficient access to constrained power and water resources, as well as data centre operators to identify emerging technologies that improve compute capacity, power delivery and cooling efficiency, as well as networking infrastructure.

Networking: A critical bottleneck for AI workloads

Large language models are too big to fit on a single GPU chip, meaning computation must be split across thousands of GPUs, requiring extensive volumes of data movement. Delays in synchronised data transfer will significantly slow down model training and inference requests, so high bandwidth, low latency networking infrastructure is critical.

At scale, this sheer volume of data transfer can also mean that traditional electronic switches and transceivers account for up to 5% of a data centre’s total energy demand (IEA, Energy and AI Report, 2023).

The inefficiency of traditional electronic switches is growing into a major obstacle, both for AI model performance and for energy sustainability.

Salience Labs: An optical circuit switch solution

Salience Labs is a leader in photonic chip solutions targeting connectivity for AI data centre infrastructure. Backed by over a decade of research from University of Oxford, Salience’s innovative developments in photonic switching technology enable high-speed, ultra-low latency networking fabrics that remove infrastructure bottlenecks for AI workloads.

Their optical circuit switches (OCS) move light instead of electrons, reducing power, increasing bandwidth, and lowering latency, eliminating unnecessary optical-electrical-optical conversions.

Salience Labs’ technology, when deployed in hyperscale AI datacentres, reduces power consumption considerably compared to traditional electronic switches and transceiver combinations and enables lower latency run times for AI models.

By enabling high-bandwidth, low-latency connectivity, Salience’s photonic switches remove bottlenecks in an AI network which unlocks more efficient AI model training and deployment, delivering significant performance gains and meaningful sustainability improvements.

Why this matters for Real Asset investors

At Taronga Group, we believe that the future of high-performing Real Assets is dependent on owners, operators and developers adopting emerging technologies. Data centres sit squarely within this intersection – operationally intensive Real Assets dependent on next-generation systems for continually improving power, cooling, computation and connectivity performance.

Taronga Groups’s investment in Salience Labs helps to address the pressing networking efficiency challenge in our global AI infrastructure build out, enabling our partners to:

• Gain deeper insight into the operational challenges within their data centre assets
• Introduce solutions that directly reduce energy use and improve sustainability
• Introduce solutions that enable their data centre operating partners to maximise the performance of their infrastructure investments

Looking ahead

The investment reinforces our role in advancing the technologies at the intersection of Real Assets and Digital Infrastructure.

As the integration between Real Assets and advanced technologies continue to deepen, Taronga Group remains focused on ensuring our partners are prepared – not just for today’s demands, but for tomorrow’s opportunities.

To find out more about Salience Labs, visit https://www.saliencelabs.ai/.