When Zerodha co-founder Nikhil Kamath put ₹200 crore into Indian data centre operator CtrlS, his reasoning was blunt: every meaningful tech shift runs on infrastructure first, applications second. It's a line UK founders should sit with, because the same capital rush is happening on their doorstep — and it's already showing up on their cloud bills.
Here's the answer up front: data centre investment is surging because AI workloads need ten times the compute of traditional software, and that scarcity is driving up colocation and cloud prices across the UK. Businesses that understand this shift now can lock in capacity and pricing before the squeeze gets worse.
What is the Concept
A data centre is the physical backbone behind every app, AI model and website a business runs — racks of servers, cooling systems and network links housed in dedicated facilities. Investors like Kamath aren't betting on software; they're betting on the physical capacity that all software, especially AI, depends on.
In the UK, this capacity sits with hyperscalers (AWS, Microsoft Azure, Google Cloud) and colocation operators clustered around London, Slough and Manchester. When global capital pours into data centre operators, it signals that compute — not code — is becoming the scarce resource businesses need to plan around.
Why It Matters in United Kingdom (2025–2026 Context)
The UK is already grappling with a well-documented data centre power problem. West London, home to one of Europe's densest data centre clusters, has faced National Grid connection delays stretching into the 2030s for new facilities. That constraint is pushing up the cost of colocation and cloud compute for every business relying on UK-based infrastructure, from a Manchester SaaS startup to a London fintech scaling its AI features.
Government moves to designate data centres as Critical National Infrastructure in 2024 have accelerated planning approvals, but demand from AI workloads is outpacing supply. For UK SMEs, this means rising GBP cloud costs are not a temporary blip — they're structural, and budgeting for 2026 needs to account for it.
How AI Is Changing This
AI training and inference are far more compute-hungry than the web apps most UK businesses budgeted for a few years ago. A single fine-tuned model or RAG pipeline can multiply a company's monthly compute spend several times over, which is exactly the demand driving investors toward companies like CtrlS globally and their UK equivalents locally.
This is where a contrarian insight matters: most UK founders assume AI costs are a software problem to be solved with a cheaper model or vendor. In reality, it's increasingly an infrastructure procurement problem — the businesses winning on AI cost efficiency in 2026 are the ones treating compute capacity like a supply chain to be negotiated, reserved and hedged, not a utility to be metered and forgotten.
Real-World Examples
Equinix and Digital Realty, both major operators in the London and Slough data centre corridor, have reported waiting lists for new capacity as AI-driven demand collides with grid constraints — mirroring the exact dynamic Kamath's CtrlS investment reflects in India. UK cloud-native firms report their AWS and Azure bills rising even without adding new workloads, simply because premium compute regions are getting more expensive as capacity tightens.
A London-based logistics-tech company, for example, found that shifting non-latency-sensitive AI inference jobs to lower-demand UK regions and off-peak scheduling cut its monthly cloud spend by close to 20% — without touching its product roadmap. That's the kind of infrastructure-first thinking the CtrlS investment signals is now table stakes globally.
Practical Insights / Actions
The founder mistake to avoid: treating cloud infrastructure as a fixed cost you review once a year. In a market where capacity is scarce, prices and availability shift quarterly, and businesses that don't actively manage compute commitments end up overpaying or getting locked into premium-only regions.
Use what we call the Compute Leverage Framework: (1) audit which workloads genuinely need premium, low-latency capacity versus which can run on cheaper, off-peak or regional infrastructure; (2) negotiate reserved capacity or committed-use discounts before demand spikes further in 2026; (3) monitor for 'compute debt' — the hidden cost of over-provisioned or idle infrastructure quietly accumulating on the bill, much like technical debt accumulates in code.
Future Outlook
Expect UK data centre and cloud pricing to stay under upward pressure through 2026 as AI adoption grows faster than grid and construction capacity can keep up. The hidden opportunity here is for UK businesses that plan ahead: securing capacity or optimising workloads now, while competitors are still treating compute as an afterthought, creates a genuine cost advantage as prices climb.
Investors backing data centre operators worldwide, from CtrlS in India to UK colocation players, are essentially pricing in years of sustained AI-driven demand. UK founders who plan their infrastructure strategy with the same long horizon will be the ones scaling profitably rather than absorbing surprise cost increases each quarter.
Conclusion
Nikhil Kamath's ₹200 crore bet on CtrlS is a reminder that every tech shift, including the UK's AI adoption wave, ultimately runs on physical infrastructure and its cost. UK businesses that audit their compute usage, negotiate ahead of demand, and treat infrastructure as a strategic line item — not a background utility — will control their 2026 costs instead of being controlled by them. RP SoftTech works with UK businesses to audit cloud and AI infrastructure spend and build leaner, scalable technology strategies; if rising compute costs are already showing up on your bills, that's the conversation worth having next.