Integrated in Coral deployments today
- Single-phase immersion with dielectric fluids
- Direct-to-chip liquid cooling
- Closed-loop dry heat rejection
- High-density GPU infrastructure
- Infrastructure monitoring and telemetry
- Private AI orchestration
Research
Technology radar
Where CoralDC places specific technologies on a readiness horizon, and what would have to change for each to move. Four horizons: integrated in deployments today, entering commercial deployment, technically credible but commercially unproven, and tracked on a long horizon. The radar is generated from a single data file and carries its review date so an abandoned radar is visible as one.
Review Cycle
Last reviewed 20 August 2026.
Reviewed quarterly. Next review 20 November 2026.
The radar is generated from a single data file, so the horizons and the review date are edited in one place and can never drift apart on the page.
The Horizons
Four readiness horizons.
Where CoralDC places specific technologies today, and what would have to change for each to move.
Entering commercial deployment, under active evaluation
- Hybrid direct-to-chip plus dielectric bath architecture
- Automated thermal control
- Continuous fluid condition monitoring
- Distributed fibre optic thermal sensing
- Waterless and dew-point heat rejection for hot climates
- Negative-pressure water loops for regulated buildings
- Per-pod digital twins
- Immersion-cooled storage integrated with the pod thermal architecture
Technically credible, commercially unproven
- PFAS-free two-phase immersion
- Grid-interactive compute at fleet scale
- Heat offtake as a standardised commercial product
- Compact additively manufactured thermal cores
- Passive thermal transport for unstaffed sites
- Heat-driven water recovery
Tracked, long horizon
- Microchannel and in-silicon microfluidics at merchant availability
- High-conductivity spreader materials
- Superconducting power delivery
- Co-located generation including microreactors
- Heat-to-power conversion at deployment scale
A View, Dated
What CoralDC expects to become ordinary
A view, offered as a view rather than a forecast, and dated so it can be judged later.
Single-phase water direct-to-chip is becoming the default and will stop being a differentiator, at which point facility water rather than chip temperature becomes the constraint. Automated cooling control will move from novelty to expectation, and within a few years selling infrastructure without it will look like selling a building without a management system. Waterless two-phase refrigerant cooling will take the high-density segment, driven by water permitting rather than by thermal performance. Interconnection rather than silicon supply will become the binding constraint on AI capacity.
And immersion as a standalone product category does not survive. It becomes a niche, or it is absorbed into hybrid architectures. CoralDC has built its roadmap on the second outcome.
Next Step
Talk to CoralDC engineering.
If you have data, a disagreement or a thesis on any of this work, CoralDC would rather hear it than not.