AI Investments Fuel Australia’s Data Centre Future

This story features GOODMAN GROUP, and other companies. For more info SHARE ANALYSIS: GMG

The company is included in ASX20, ASX50, ASX100, ASX200, ASX300 and ALL-ORDS

Updated sector analysis and forecasts suggest fears of the demise of data centre stocks and the GenAI revolution are overblown.

-BigTech’s March quarter results dismiss investors’ AI oversupply concern
-Analysts dissect the demand/supply dynamic for data centres
-APAC data centre growth outpaces global growth
-Australia a geopolitically attractive destination for data centres
-Local favourites to invest in exposure

By Danielle Ecuyer

DeepSeek R1 and the Infrastructure Reset

The durability of the AI trade and the capacity of existing infrastructure, particularly data centres, to support generative AI, such as large language models, were tested on January 27, 2025, when China’s DeepSeek unveiled its R1 model: an open-source LLM that matched or outperformed leading Western models like OpenAI’s GPT-4 at a fraction of the cost.

Reportedly developed using less than US$6m in computing resources, R1 challenged the prevailing belief that state-of-the-art AI demands massive capital and high-end infrastructure, such as super-sized data centres.

Its release triggered a sharp sell-off in tech stocks, with Nvidia suffering a historic single-day loss of -US$593bn in market capitalisation. Other major tech firms, including Microsoft, Alphabet, and ASML, also experienced significant declines.

Concerns about data centre infrastructure spending, combined with equity raisings from Goodman Group ((GMG)), several high-profile transactions, and the listing of DigiCo REIT ((DGT)), prompted profit-taking in Australian data centre companies, with the narrative shifting to risks around demand and potential overcapacity in the sector.

Market sentiment rebounds post-March quarter earnings

By May, with considerable developments reshaping the AI landscape including rising tariff and trade tensions between the US and China, the major US tech companies through their latest quarterly earnings have, for now, eased market concerns.

The Big Tech March quarter earnings reports were on balance better than feared and, homing in on the specific results from the hyperscalers, fears over reduced investment spending, growth in GenAI and associated cloud-based infrastructure proved to be ill-founded.

UBS estimates the latest capital spending as highlighted in the updates from Microsoft (Azure), Amazon (AWS), Meta and Google (Google Cloud Services) is US$330bn in 2025, a rise of 34% on 2024.

Morgan Stanley noted Microsoft reported one of the strongest quarters in “recent memory” with Azure growth increasing to 35% on a year previously, well above guidance at 31%-32% and an acceleration over the 31% growth in the previous quarter.

The analyst estimates Azure AI growth has risen to over 215% year-on-year and the company flagged ongoing issues dealing with supply versus demand imbalances post the June quarter.

Meta offered a robust result pointing to AI-driven feed and recommendations underpinning improved engagement.

While Amazon’s AWS is spending up on more GenAI efficiency enablers, coding, and sales assistants to improve teams’ efficiency with the same or fewer headcount.

AI mentions accelerate in 2025 reporting season

Checking in with ChatGPT on the latest updates on GenAI by companies globally, the service offered the following:

AI Mentions Surge in 2025 as Companies Ramp Up Deployment

In 2025, a dramatic rise in corporate adoption and disclosure of artificial intelligence has reshaped how businesses operate and communicate with investors.

According to recent industry analyses, over 78% of global companies now use AI in at least one part of their operations, with adoption accelerating across IT, customer service, and marketing functions.

The impact is particularly visible in financial reporting.

More than 320 companies referenced AI in their annual reports this year; a 152% increase from 2024.

Among Fortune 500 firms, over half (281 companies) now cite AI not only as an operational tool but also as a material risk factor, up from just 49 the previous year.

This momentum is being driven by a sharp rise in generative AI usage, which has more than doubled year-on-year, with 71% of organisations employing such technologies in 2024.

Industry leaders in finance, healthcare, manufacturing, and retail are integrating AI to boost productivity, automate workflows, and enhance personalisation.

While AI’s benefits are widely touted, the shift also raises questions around data ethics, transparency, and displacement.

Still, corporate disclosures suggest the technology has become integral to strategic planning, no longer a fringe innovation but a mainstream engine of growth.

Australian data centre stocks rebound

Against this backdrop, listed Australian data centre companies have seen share prices bounce off the Trump Tariff Tantrum lows, but all remain well below the pre-DeepSeek highs.

Citi and UBS are two brokers which have taken the challenge head-on to assess the data centre industry, noting multiple drivers of demand.

Citi’s research focused on the demand for chips and the potential for oversupply, which leads back to establishing a baseline for data centre demand.

The broker points to data centre demand more than cloud revenue growth in 2023, as companies sought to firm up leases in data centres ahead of emerging GenAI trends and workloads.

Data centres were confirmed as the main beneficiaries of expanding demand for GenAI workloads, and these workloads are estimated to represent more than 50% of data centre power demand by 2030.

Sizing the market: Co-Location vs Hyperscale

Depending on the criteria measuring data centre demand, i.e. on-premise enterprise versus co-location (when companies lease space in data centres), the expected growth rates vary. Citi estimates 17% compound average growth rate (CAGR) through to 2030 with co-location currently representing around 33GW.

Historically, co-location is just under a quarter of global data centre capacity, with hyperscalers commanding around 40% and on-premise enterprise data centres representing just over 35% in 2024. The trend has been away from on-premise to more scalable, cost-effective solutions which co-location and hyperscalers offer.

APAC capacity accelerates

UBS points to APAC data centre capacity advancing by 27% CAGR over the past five years, more than global growth at 17.3% over the same period.

China represents 64% of capacity, followed by Japan, Australia and India, each with 5%. APAC ex-China grew at a 19% CAGR between 2019-2024, while China’s capacity grew at 29% CAGR.

APAC data centre capacity is estimated at 24MW, circa 40% of global capacity.

UBS estimates a CAGR for demand over 20% between 2024-2030 for the APAC region due to higher rates of cloud adoption and migration overlaid with an AI multiplier of over two times.

Drivers of Demand

Picking apart the demand drivers specifically, growth in subscriber numbers for new technologies like 5G, Internet-of-Things (IoT), and streaming all require increased levels of storage at data centres.

Migration to the cloud for both hardware and software cloud computing is opposed to on-premise facilities.

The UBS US enterprise software analysis estimates global cloud infrastructure penetration at around 35%, which is about 50% through the penetration curve.

The ceiling is estimated to be capped at circa 70% due to expected government restrictions and mainframe workloads which stop 100% cloud migration.

Cloud infrastructure revenue is growing between 15%-20% per annum and is likely to accelerate with AI growth and adoption.

Applying AI is expected to increase the CAGR for APAC demand to 41% compared to 12% without AI out to 2030, equaling an AI multiplier by FY30 of 2.8 times, against 1.3 times in FY25.

The analyst quotes an Australian data centre contact who stated, “the table stakes for AI is now over 75MW” and average workloads with AI are much higher.

Across the APAC region, UBS analysed the probability of a supply glut, determining Singapore has zero percent with China the highest probability of oversupply at 19.8%.

On relative valuation and supply glut basis, Singapore, Malaysia and South Korea seem most appealing, while Australia is ascribed an 8% probability, low against the others, just relatively less attractive on an estimated cash flow return on investment.

Overall, a 16% probability of an APAC data centre glut is estimated based on demand growth of 19% p.a. and supply growth of 17%-18% annually.

Supply constraints: power and permits

While markets have tended to focus on demand, challenges to supply and potential expansion constraints are equally important.

At Macquarie’s recent Asia Conference, the speakers on the data centre panel coalesced around the views that:

-power remains the largest constraint to speedier data centre development,

-concerns around demand waning are overdone with “compute supply” challenged to keep pace.

The panel notably highlighted supply chain issues resulting from restrictions on GPU sales to ASEAN markets and Trump tariffs which have created uncertainty and market volatility.

Regarding power, bottlenecks for faster data centres developments have pushed out lead times to between five and seven years from two to three years around three years ago in the US.

Combining co-location gas turbines generation and gas connections with data centres is also experiencing more elongated time frames.

Power infrastructure bottlenecks in APAC

In the APAC market, the power procurement process is dependent on the structure of specific markets.

UBS observes total capacity in APAC cities, including Tokyo, Sydney, Singapore, Johor, Seoul and Mumbai, has risen 59% since the end of 2022 by 3.8GW, which is due to an increase in planned capacity. Most is referred to as pipeline capacity, i.e. under construction and planned.

The actual increase in supply stands at 1.6GW or 36% growth against two years ago, lower than the expected 81% growth in pipeline capacity.

Again, power availability and the power grid are pointed to as a major bottleneck which is heightened by the fact data centres consume around ten times the energy per floor space of a usual office building.

Computing power and servers represent 40% of power usage, with cooling systems at another 40%, and the balance comes from network equipment, storage systems, etc.

Australian data centre power usage is estimated at 2% and going to 5% of power supply, compared to the average for ASEAN at 2.1%. Singapore is the highest at 7%, going to 9%.

Australia is not expected to experience tight power infrastructure (power supply and grid capacity). In contrast, Japan and India are most likely to be the tightest.

Australia’s growth outlook in data centre capacity

Focusing specifically on Australia, data centre capacity has expanded at a CAGR of 20% p.a. to FY25 from FY20 and is forecast to expand out to FY30 at an 18% p.a. CAGR.

AI penetration is estimated to be less than 10% of workloads in Australia and should advance to over 50% by FY30.

UBS points to Goodman Group ((GMG)), NextDC ((NXT)), DigiCo REIT ((DGT)), and Infratil ((IFT)) experiencing rising size and volume workloads. Australia and NZ are viewed geopolitically as attractive destinations.

Citi’s analysis has around 1GW of total live capacity for Australia currently, with Sydney the largest market and in the top 20 globally at around 670MWs of live capacity.

Sydney’s capacity has expanded 2.5 times in the last five years while Melbourne’s capacity has doubled over the same period. Conversations with the industry suggest there is little capacity available in both markets.

One of the main drivers of demand has been the Federal Government’s digital and cloud-first strategy, including data sovereignty.

Applying Citi’s expectations of Australia to grow in line with the APAC region at CAGR of 14% to 2030, the implied data centre demand will grow to 2.5GWs.

NextDC: Engineering differentiation in AI

NextDC is the second-largest data centre provider in Melbourne after AirTrunk and the fourth-largest provider in Sydney, after AirTrunk, CDC and Equinix.

UBS believes the structural demand for NextDC remains robust via the shift to the cloud, growth in co-location, and AI-related demand.

The company is highlighted as one of Australia’s top co-location, carrier-neutral data centres with a “deep competitive moat” due to the large ecosystem of enterprise customers.

NextDC has 17 operational data centres with 10 in planning/development and six sites being evaluated. Expansion into Japan, Thailand and Malaysia positions NextDC as an active APAC participant.

UBS is Buy-rated with a $19.80 target price.

Citi views NextDC as the best direct exposure for the AI thematic in APAC as the company’s offering is suitable for both hyperscale and enterprise customers.

The company’s engineering skills are also highlighted as a key “differentiator”. September’s equity raising was observed by the analyst as sooner than expected but does remove additional equity funding for the next three years.

The latest 50MW contract win was in line with expectations and commentary suggested the contract will achieve the top revenue run-rate by FY28. The deal was noted for being the first AI-related contract.

Citi has a Buy rating with a $18.70 target price. FNArena’s consensus price target is $19.50 with six Buy-equivalent ratings, including Citi and UBS.

Goldman Sachs is also Buy-rated, with its target price set at $16.50.

Infratil’s CDC Stake: Strategic and growing

NZ-headquartered Infratil ((IFT)) has a 49.75% stake in CDC Data Centres, which UBS estimates is worth NZ$5bn and is one of the largest operators of data centres with a 24% market share in 2024.

Infratil is Buy-rated with a NZ$14 target price.

Daily monitored broker Morgan Stanley also has a Buy-equivalent rating with a NZ$15 target price.

Goodman Group: Repricing the pipeline

Goodman Group is also on UBS’ radar with increased confidence in the group’s data centre funding outlook, highlighting Goodman as unique for a REIT with around 10% medium-term operating EPS growth (forecast).

While noting the $4bn equity raising which came as a surprise, the market was viewed as too quick to discount the group’s data centre pipeline of 5GW at $100bn in the preceding months without considering funding.

UBS estimates over $10bn in capital by FY29 to commence 5GW with joint venture agreements of circa $6bn, with the latter able to be funded from internal cashflow, addressing concerns that more capital would need to be raised.

Goodman Group is Buy-rated with a $36 target price.

Morgan Stanley made some very prescient calculations on Goodman Group at the start of equity market’s tariff tantrum.

The analyst valued the group’s industrial warehouse business at $24.94bn, including work in progress of $8bn, which would underpin around 9% compound average growth rate in EPS and generate circa 70% of FY26 EPS forecast.

At that point the market was undervaluing the data centre business and the powerbank versus global data centre peers.

FNArena’s daily monitored brokers have a consensus target price of $36.435 with five Buy-equivalent ratings and one Hold-equivalent rating.

DigiCo and Megaport: Smaller names with AI leverage

Turning to DigiCo with 13 assets across Australia and the US, UBS believes the REIT is positioned well to leverage existing sites such as SYD1. DigiCo is viewed as offering an attractive growth outlook with an expected five-year CAGR for earnings (EBITDA) of over 23%. Buy-rated with a $5.60 target price.

Citi also points to Megaport ((MP1)) as a beneficiary of GenAI and ongoing cloud adoption with its data centre and cloud-on-ramp presence globally. The analyst proposes inference will be a bigger driver of workload demand for Megaport.

Megaport is Buy-rated with a $9 target.

FNArena’s consensus target price sits at $11.21 with four Buy-equivalent ratings and two Hold-equivalent ratings.

FNArena has a dedicated GenAI section at https://fnarena.com/index.php/tag/gen-ai/ including the following stories

https://fnarena.com/index.php/2025/05/09/dug-technologys-multiple-growth-levers/

https://fnarena.com/index.php/2025/04/03/megaports-quest-of-data-centre-agnostic-growth/

https://fnarena.com/index.php/2025/03/19/the-genai-arms-race-is-just-getting-started/

Find out why FNArena subscribers like the service so much: “Your Feedback (Thank You)” – Warning this story contains unashamedly positive feedback on the service provided.

FNArena is proud about its track record and past achievements: Ten Years On

Click to view our Glossary of Financial Terms