Every agent that completes a task, every model that answers a question, every robot that moves through a warehouse depends on a chip. As AI moves from experiment to infrastructure, semiconductors have become the foundation of economic and national competitiveness.
Demand is structural
AI workloads require accelerators, high-bandwidth memory, advanced networking silicon and power semiconductors at a scale the industry has never supplied. Inference — running models in production, continuously, for billions of users and agents — is set to become the dominant source of demand, and it rewards specialised, efficient silicon.
Supply is strategic
Leading-edge manufacturing, advanced packaging and critical materials remain highly concentrated. Governments across allied economies are investing to diversify supply chains, creating a generational opportunity for private capital to partner in the build-out.
Compute is the new oil, and silicon is the refinery.
Where we focus
- Custom and inference silicon: domain-specific chips that dramatically improve performance per watt.
- Advanced packaging: chiplets and memory integration, where much of the next decade's performance gain will come from.
- Equipment and materials: the tools, substrates and specialty materials every fab depends on.
- Power semiconductors: silicon carbide and gallium nitride devices for data centres, EVs and the grid.
Beyond classical computing
We also invest selectively in quantum computing — particularly enabling hardware such as cryogenics, photonics and control electronics, and post-quantum security — where nearer-term revenue meets long-term optionality.
At Blacksaila, semiconductors sit at the centre of our AI stack: below the agents and applications we own, alongside the compute we build, and above the power we invest in.