Why Are Billionaire Investors Like Nikhil Kamath Betting Big on Data Centres in 2026 — What Does It Mean for Australian Businesses?

When Zerodha co-founder Nikhil Kamath quietly wrote a ₹200 crore cheque (roughly AUD $36 million) into Indian data centre operator CtrlS, he wasn't chasing a hot IPO or a flashy AI app — he was betting on the pipes, not the water. His line, 'every meaningful tech shift runs on infrastructure,' is a warning most Australian founders are ignoring: while everyone argues over which AI model to use, the smart money is quietly buying the racks, power contracts and cooling systems those models actually run on, and Australia's own compute capacity race is about to feel that pressure directly.

What is the Concept

A data centre business isn't a tech company in the way most people picture one — it's a real-estate, energy and networking company that happens to host servers. CtrlS, like NextDC in Australia, sells colocation space, power and connectivity to banks, cloud providers and AI labs who don't want to build their own facilities. When an investor like Kamath puts capital into that layer rather than into an app or a model, he's making a structural bet: every AI chatbot, SaaS dashboard and automation tool an Australian business uses this year ultimately depends on a physical building somewhere with enough power and cooling to keep GPUs running.

This is worth naming properly. Call it the Infrastructure Gravity Principle: in every major tech shift — cloud in the 2010s, AI in the 2020s — capital gravitates first toward the scarcest, least flexible layer of the stack, which is physical compute infrastructure, not the software built on top of it. Kamath's CtrlS bet is a textbook example, and it's a signal Australian operators, CFOs and technology leads should be reading closely rather than dismissing as an India-only story.

Why It Matters in Australia (2025–2026 Context)

Australia is already living this shift. Sydney and Melbourne remain the country's primary data centre hubs, but capacity constraints — grid connection delays, land scarcity, and rising electricity prices — are pushing new development toward Western Sydney, outer Melbourne, and increasingly Perth, where access to renewable power is attracting hyperscale interest. Under the Security of Critical Infrastructure Act and the Privacy Act, Australian businesses handling sensitive data also face real pressure to host workloads locally rather than defaulting to offshore cloud regions, which only intensifies domestic demand for data centre space.

The dollar figures make the point. AirTrunk, an Australian-founded data centre operator, was acquired by Blackstone in 2024 for roughly AUD $24 billion — one of the largest infrastructure deals in Australian corporate history. NextDC, listed on the ASX, continues to expand its Sydney (S5) and Melbourne (M3) campuses to meet enterprise and AI demand. Kamath's ₹200 crore move into CtrlS is small by comparison, but it confirms the same global thesis institutional capital is already acting on in Australia: compute infrastructure is now treated as essential utility-grade real estate, not a niche IT expense.

How AI Is Changing This

Traditional data centres were built for steady, predictable loads. AI workloads are different — GPU clusters draw far more power per rack, generate more heat, and often need liquid cooling instead of standard air conditioning. This is forcing Australian operators to retrofit or build new 'AI-ready' facilities, and it's a key reason colocation pricing has climbed faster than general inflation over the past two years.

There's also a sovereign AI angle developing locally. As Australian government agencies and regulated industries (banking, healthcare, defence-adjacent sectors) push to keep AI training and inference data onshore, demand is shifting toward Australian-owned or Australian-hosted capacity rather than relying purely on US hyperscale regions. That mirrors exactly the logic behind Kamath's CtrlS bet — capital following AI demand back down to the physical layer, in whichever country needs sovereign compute most urgently.

Real-World Examples

AirTrunk's $24 billion sale to Blackstone in 2024 remains the clearest Australian proof point that global institutional capital sees data centres as core infrastructure, not speculative tech. NextDC's continued expansion across Sydney, Melbourne, Brisbane and Perth, and Macquarie Data Centres' government and enterprise contracts in Canberra, reinforce the same pattern domestically: steady, long-term capital chasing compute capacity rather than short-term app hype.

CtrlS itself is instructive too. It has been expanding beyond India into other Asia-Pacific markets, and given growing India-Australia trade and technology ties, Indian data centre operators eyeing regional expansion is a realistic scenario worth watching — not a hypothetical. Kamath's investment isn't an isolated India story; it's one data point in a broader APAC infrastructure buildout that Australian businesses are already inside of, whether they've noticed or not.

Practical Insights / Actions

The common founder mistake in Australia right now is waiting for cloud and compute prices to come down before committing to infrastructure decisions. That bet is backwards — capacity is tightening, not loosening, as AI demand grows faster than new data centres can be built and connected to the grid. The hidden opportunity is in locking in colocation or hybrid-cloud arrangements early, and in exploring state and federal incentives tied to sovereign AI and digital infrastructure investment before they become oversubscribed.

Practically, Australian SMEs and mid-market companies should start with an honest audit: which workloads genuinely need premium, low-latency local infrastructure, and which can run on standard public cloud. This is exactly the kind of infrastructure and AI-readiness assessment RP SoftTech works through with clients — mapping current compute costs against future AI workload needs so infrastructure decisions are made proactively, not in a scramble once prices rise further.

Future Outlook

Expect data centre capacity in Australia to keep expanding through 2026 and beyond, but not fast enough to fully offset AI-driven demand — meaning colocation and cloud compute pricing is likely to stay elevated in the near term, particularly for GPU-intensive workloads. Grid connection timelines and renewable energy availability will increasingly determine which cities win new investment, with Perth and regional Victoria positioned to gain ground on Sydney and Melbourne.

The strong opinion worth stating plainly: Australian businesses that keep treating compute infrastructure as a background utility bill will lose ground to competitors who treat it as a strategic asset to be planned, negotiated and secured years in advance — exactly the mindset Kamath's CtrlS bet represents at institutional scale.

Conclusion

Nikhil Kamath's ₹200 crore CtrlS investment is a small transaction with a large lesson: every meaningful tech shift, AI included, runs on physical infrastructure before it runs on software. For Australian founders and operators, the takeaway isn't to chase data centre stocks — it's to audit compute dependency now, before capacity constraints and rising prices force a reactive decision. If you're unsure where your business stands, RP SoftTech offers infrastructure and AI-readiness audits to help Australian companies plan this shift with clarity instead of guesswork.