Should you invest in data centre stocks? Here are 4 reasons why you should consider, and 4 reasons why you should not

Nick Sundich Nick Sundich, January 28, 2026

If you want to invest in data centre stocks, we would assume you are most likely thinking that it is a no-brainer as long as you have patience. You could just simply say: NextDC (ASX:NXT), alluding to that company’s 100-fold growth from an $80m start-up of Bevan Slattery’s to an $8bn company. You may or may not be thinking of the environmental impact of data centres and likely decided that the growth outweighs any concerns or that any impacts will be offset by the technologies the centres power. But the investment case for data centre stocks is much more complicated than that.

Let’s look at 4 reasons to consider data centre stocks and 4 reasons to not consider them.

What are the Best ASX stocks to invest in right now?

Check our buy/sell tips

4 reasons you might invest in data centre stocks

1. The demand for data – its legitimate

Investing in data centres and data-centre companies can make a lot of sense right now because they sit underneath some of the most powerful structural trends in the global economy. Demand for cloud computing, AI training and inference, streaming, gaming, fintech, and enterprise software all ultimately turns into demand for physical compute and storage, and that compute has to live somewhere. Macquarie estimates that data centre power capacity grew from 26 gigawatts (GW which is worth a billion watts) in 2015 to 81Gw in 2024, a more than tripling in that time. This reflects not only more facilities being built but each facility drawing more power as workloads intensify.

Now, investors may think that the cloud can just store everything and as capable as the cloud is, there is still a place for data centres. Training large models and running inference at scale requires far more power density, specialised cooling, and proximity to reliable energy than what the cloud can provide.

2. Data centres are pick and shovel plays

If companies like Microsoft, Amazon, Google, or Nvidia keep spending aggressively on AI infrastructure, data-centre operators effectively become picks-and-shovels providers to that arms race. Some investors may have heard the term ‘pick and shovel’ before, but others may have not. This term emerged from the 19th century gold rushes when the biggest winners were not necessarily the mining companies (although some won) but the companies that sold picks and shovels. They captures value from the activity of mining rather than the outcome.

Whether the future belongs to OpenAI, Google, Anthropic, open-source models, or players not yet imagined, all of them require massive amounts of compute, storage, networking, and power. Data-centre operators monetise the existence of that demand without needing to guess the ultimate winners.

3. Operators have strong pricing power

You might be saying at this point,’ OK, data centres make a lot of revenues but so do supermarkets and they are low margin’.  Data centres tend to benefit from long-term contracts with large, creditworthy customers like hyper scalers and governments, which can make revenues relatively predictable. In many markets they also have strong pricing power, because capacity is scarce, planning approvals are slow, and building new facilities takes years and huge capital.

Once a data centre is built and leased, switching costs for customers are high, so churn tends to be low. For listed data-centre companies and REITs, this can translate into steady cash flows, inflation-linked pricing in some contracts, and exposure to secular growth rather than cyclical consumer demand.

4. The resilience of demand – for macroeconomic and monopolistic/oligopolistic reasons

Even in economic slowdowns, enterprises and governments are unlikely to turn off cloud services, data storage, or core digital systems, which helps protect occupancy and revenue. In key markets, limited power availability and zoning restrictions create high barriers to entry, supporting stable rents and making existing assets more valuable over time.

Just look at Equinix, one of the largest data centre players. In recent quarters the company has been able to raise prices on thousands of customer contracts globally, generating material incremental revenue even as energy costs rise — a clear indication that customers accept those increases because the service is essential and hard to replace. For example, Equinix reported raising pricing on more than 7,000 customers across 16 countries, adding roughly $90 million of incremental revenue in a single quarter just by adjusting prices rather than signing net new customers. Of course, being a major market players also helps.

4 reasons you might not invest in data centre stocks

1. The Capital intensity of data centres

Data centrexs require massive upfront investment in land, power connections, cooling, and hardware, and returns depend heavily on keeping utilisation high. If supply gets ahead of demand in a region, pricing can fall quickly and margins can compress. Miami and London-based corporate advisory shop CleanBridge reckons that a typical data centre can cost roughly $7–$12m per MW of IT load capacity just to build the facility itself — land, shell, electrical and cooling infrastructure included. That doesn’t count the servers, GPUs, racks, storage, and networking gear, which often add several times that amount in total initial investment for a fully functioning centre.

To make that more tangible, consider real corporate development pipelines: Digital Realty, one of the world’s largest data-centre owners, had a pipeline of projects totalling about $3.86bn in development costs, representing 321 MW of power capacity and millions of square feet under construction — implying roughly $12m per MW of power capacity built. Equinix, another major operator, had roughly $952m of ongoing hyperscale builds totalling about 83 MW, implying a cost of ~$11.5m per MW.

2. Power and water costs can be enormous

Electricity costs are one of the largest operating expenses, and constraints on grid capacity or rising energy prices can materially hurt profitability or delay expansion. Specifically, Australia’s energy market operator estimates data centres consumed 4 terawatt-hours of electricity, or 2% of total powder delivered in the national grid. At current rates, this will be 21.4 terawatt-hours and 9% of the total. Of course this is only one estimate – McKinsey reckons it could be 13% of the grid by 2030 if ambitions to make Australia a regional hub are realised. Investors may say that data centres can just continue to raise their prices and the grid will just keep producing. But this may not be so, as we’ll get to below.

3. Regulation and politics are an increasing concern

Concerns about energy use, water consumption (needed to cool data centres), and carbon emissions are already leading to stricter rules in some jurisdictions, which can raise costs or cap growth. This is not just because of public pressure that politicians worried about their jobs are responding to. The loads of data centres (equivalent to small cities) are causing problems.

In parts of Ireland, for example, authorities have effectively restricted new data-centre grid connections around Dublin because centres were projected to consume a large share of national electricity demand, raising concerns about energy security and household prices. In parts of southern Europe there has been stricter permitting, mandatory water-use disclosures, or requirements to adopt more expensive closed-loop or liquid-cooling technologies.

On carbon emissions, governments are increasingly unwilling to allow large new sources of electricity demand unless they align with climate targets. Some jurisdictions now require data-centre developers to demonstrate access to renewable energy, purchase long-term power-purchase agreements, or meet specific efficiency benchmarks before permits are granted. In markets with carbon pricing or aggressive net-zero commitments, this can materially increase costs and slow development timelines. It also introduces regulatory uncertainty: a project that looks viable today may face new constraints halfway through its development cycle.

4. Data centre stocks can be overvalued

For public investors, valuations can also be a problem—many data-centre companies are priced for near-perfect growth driven by AI and cloud. Equinex’s P/E is over 60x, far above what many investors would consider is reasonable. This suggests the market is pricing in sustained, strong earnings growth for the years ahead. But this is not matching up with reality – at least not now.

Any slowdown in demand, change in hyper-scaler strategy, or technological shift toward more efficient or decentralised computing has the potential to hit share prices hard.

So should you or should you not invest in data centres?

In our view, only if you remember the following fact. That data centres are not risk-free assets: they consume a lot of time, money, power, water and they emit a lot of carbon. Yes, they can generate a lot of money when all is said and done…but only at that point.

This may be true of all stocks, that investors should only invest if they have considered the risks. But we’d argue in no other sector

Blog Categories

Get Our Top 5 ASX Stocks for FY26

Recent Posts

Austal (ASX:ASB) Wins A$4bn Navy Deal: Is It Time to Buy the Dip?

Austal’s Defence Surge: What Investors Should Watch Austal (ASX: ASB) closed at A$6.30 on Friday, up 5.53% on the day,…

Oil Surges on Iran Crisis: Best ASX Energy Stocks to Buy Now

ASX Energy Stocks: What the Oil Spike Means Now Oil prices surged more than 5% in just two trading sessions…

ASX Reporting Season Halftime Report: 4 Stocks to Buy, Hold, or Avoid After This Week’s Results

ASX Reporting Season: What This Week’s Results Mean for Investors We are halfway through the ASX reporting season, and the…