Dominic Jainy stands at the forefront of the modern digital revolution, bringing a wealth of expertise in artificial intelligence, machine learning, and blockchain technology to the table. As an IT professional deeply concerned with the physical realities of the digital age, he specializes in the delicate intersection of high-density compute power and sustainable infrastructure. In our discussion, he addresses the urgent need for a strategic national plan for the United Kingdom, moving away from reactive planning toward a model of grid discipline and ecological accountability. We explore how to dismantle the “paper project” bottleneck, the transition to closed-loop liquid cooling, and the potential for a circular compute economy that treats waste heat not as a byproduct, but as a vital community asset.
How can the United Kingdom reform its current planning and power allocation systems to prevent speculative projects from stalling the development of genuine digital infrastructure?
The shift away from the “first-come, first-served” mentality is long overdue because it allowed developers to reserve scarce electrical capacity without showing they had the financial backing or the shovel-ready plans to actually build. We have been stuck in a pattern of “grid-squatting,” where paper projects created an artificial scarcity of high-voltage capacity, leaving viable, high-demand developments in a state of limbo. To fix this, we need to apply rigorous grid discipline through regulatory reforms that demand financial commitment checks and strict project progression milestones as we move forward through 2026. This ensures that the energy we do have available is funneled directly into facilities that contribute to national compute resilience rather than sitting idle on a spreadsheet. By filtering out these speculative entries, we can finally get a clear, honest picture of our capacity, making it easier for the energy system to breathe and adapt to real-world demands.
In the absence of clear visibility for local planning authorities, what role should a unified national spatial register play in aligning data center growth with the strength of the electrical grid?
Currently, the distribution of megawatts across the country is often a mystery to central policymakers because local planning registers don’t talk to each other in a meaningful way. We need a unified, national spatial register that maps proposed compute capacity directly against regional renewable generation and the actual physical strength of the grid. This isn’t just about a map; it is about creating a visual, data-driven strategy that tells us exactly where we can locate a data hall so that it strengthens the energy system instead of straining it. Imagine being able to see in real-time where the power is greenest and the grid is sturdiest, then incentivizing developers to build there. This kind of transparency would transform data centers from passive utility draws into active, strategic partners in our national energy infrastructure.
With public concern rising over water security in specific regions, how do modern engineering standards like closed-loop cooling change the conversation about the environmental impact of these facilities?
The anxiety surrounding water security is incredibly high, especially in the water-stressed catchments across the East and Southeast of England, but much of that fear is rooted in outdated technology. Legacy systems relied on evaporative cooling towers that essentially bled local water supplies dry, but the high-density silicon we are seeing in 2026 makes those old methods physically impractical. We are moving toward a baseline of closed-loop direct-to-chip or immersion liquid cooling, which recirculates sealed thermal fluids with near-zero operational water loss. If we mandate these closed-loop systems as a standard condition for planning approval, we can protect local water basins while giving operators the extreme thermal performance they need for AI workloads. It is about shifting the narrative from a polarized debate about “power vs. water” to a technical standard that proves we can have high-performance compute without depleting our natural resources.
How can we balance the massive carbon debt incurred during the manufacturing of specialized silicon with the need for operational energy efficiency in a circular compute economy?
This is the most complex challenge we face because the embodied carbon—the energy used to manufacture, ship, and build this infrastructure—can account for up to 50% of a facility’s total lifetime environmental impact. We have been trapped in a wasteful habit of replacing functional servers every three years just to chase small performance gains, which creates a mountain of unnecessary carbon debt. However, we cannot simply keep old hardware running forever because legacy chips consume significantly more electricity per instruction, which eventually eclipses the initial manufacturing savings as the grid continues to decarbonize. The solution is workload-matched hardware placement: we should run compute-intensive AI training on brand-new, liquid-cooled silicon where every token is optimized for efficiency, while cascading legacy enterprise applications onto refurbished, secondary hardware. By extending the lifespan of these servers to six or eight years in a blended stack, we can amortize that embodied carbon without wasting the grid’s energy on heavy, modern workloads.
Given the vast amount of thermal energy generated by server clusters, how can we transition from venting this heat into the atmosphere to utilizing it as a valuable public asset?
Every single megawatt of power that enters a server hall eventually exits as low-grade heat, and in a nation striving for net-zero, venting that energy into the sky is a massive missed opportunity. We need to stop seeing waste heat as a byproduct and start treating it as a valuable community resource under a proper heat network zoning framework. This means that if a data center is located near a hospital, a housing development, or an industrial cluster, integrating waste-heat recovery should be a standard requirement for its planning consent. Of course, we have to be smart and avoid mandating heat exports where no local “heat sink” exists, as that would just lead to stranded capital. But when the geography aligns, these facilities can act as the thermal engines for entire neighborhoods, turning the hum of the server room into the warmth of a local home.
What is your forecast for the evolution of sustainable digital infrastructure?
I believe that over the next few years, from 2026 to 2028, we will see the total integration of the data center into the urban and environmental ecosystem, moving from an isolated “black box” to a multi-functional utility. We are moving toward a future where “compute resilience” is synonymous with “ecological defense,” and the facilities that thrive will be those that offer more to the community than just processing power. We will see the rise of the “blended stack” where hardware is kept in the loop for twice as long as it is today, and the heat generated by your AI queries will be the same energy heating your local swimming pool or clinic. The speculative gold rush is ending, and in its place, we are building a more disciplined, circular, and thermally efficient backbone for the digital economy that respects the physical limits of our planet.