Zerodha co-founder Nikhil Kamath just put ₹200 crore (roughly CAD $32 million) into CtrlS Datacenters, saying 'every meaningful tech shift runs on the infrastructure nobody sees.' For businesses in Toronto, Vancouver, and Montreal, this isn't a story about India — it's a preview of the compute crunch already driving up AI adoption costs across Canada in 2026.
What is the Concept
Data centres are the physical backbone of every AI product, SaaS platform, and cloud service a business relies on. When a high-profile investor like Kamath puts serious capital into a data centre operator rather than an app or platform, it signals where the real bottleneck in the AI economy sits: raw compute and power capacity, not software.
Call this the Infrastructure-Before-Application Rule — capital moves to data centres years before it shows up in consumer-facing AI tools, because every chatbot, forecasting model, or automation platform a Canadian company adopts ultimately runs on someone's server racks. Kamath's bet on CtrlS is an early signal that this layer is where the next wave of value, and cost, will concentrate.
Why It Matters in Canada (2025–2026 Context)
Canada is already living this trend. Quebec's cheap hydroelectric power has turned Montreal and Quebec City into magnets for hyperscale data centre construction, with Microsoft, Google, and AWS all expanding regional capacity through 2025 and into 2026. Alberta and Ontario are seeing similar interest as AI workloads outgrow existing capacity.
The downstream effect for Canadian businesses is rising and volatile cloud and GPU compute pricing. SMEs in Calgary and Vancouver running AI-powered analytics or customer service tools are already seeing usage-based cloud bills climb faster than their revenue from those tools — a direct consequence of the same capacity crunch Kamath's investment is chasing.
How AI Is Changing This
AI workloads, especially large language model inference, need far denser power and cooling than traditional web hosting. This is pushing data centre operators — CtrlS included — toward liquid cooling and higher power-density designs, which cost more to build and, eventually, more to rent.
For Canadian businesses, this means the era of assuming cloud AI tools will keep getting cheaper is ending. Instead, companies are increasingly choosing between committing to reserved capacity at Canadian providers or accepting price volatility from on-demand cloud AI services.
Real-World Examples
Quebec-based QScale is building gigawatt-scale, hydro-powered data centre campuses specifically to host AI training workloads, while Montreal's eStruxture has expanded capacity nationally to meet enterprise AI demand. Both are Canadian answers to the same capital shift CtrlS represents in India.
Kamath's ₹200 crore move mirrors what's happening domestically: investors and hyperscalers are racing to control physical compute capacity before demand outstrips supply. A Canadian logistics firm renting GPU-backed forecasting tools today is competing for the same finite server capacity as global AI labs.
Practical Insights / Actions
The most common founder mistake in Canada right now is treating AI tool subscriptions as fixed costs. They aren't — usage-based AI pricing is directly exposed to the same infrastructure crunch Kamath is investing behind, and bills can spike without warning as underlying compute gets scarcer.
Businesses should audit which AI and cloud workloads are cost-sensitive versus mission-critical, and negotiate reserved-capacity pricing where volume justifies it. RP SoftTech works with Canadian businesses to architect cost-efficient AI and cloud infrastructure strategies — right-sizing compute spend before it becomes an unplanned cost centre.
Future Outlook
Expect capital like Kamath's ₹200 crore bet to keep flowing globally into data centre operators through 2026 and 2027, with Canada's hydro-rich provinces capturing an outsized share of new hyperscale builds. Compute will likely stay a strategic, budgeted line item for Canadian businesses rather than a background utility cost.
Conclusion
Kamath's contrarian bet isn't really about CtrlS — it's a bet that infrastructure, not applications, will decide who wins the next phase of AI adoption. Canadian businesses that plan their compute strategy now, rather than reacting to rising bills later, will be the ones who scale AI profitably. Talk to RP SoftTech for a free AI infrastructure cost audit tailored to your business.