Investors may need to exercise patience before seeing tangible earnings emerge from the AI boom, a portfolio manager has highlighted.
Speaking at a Bennelong Funds Management media briefing last week, Sarah Shaw, chief investment officer at 4D Infrastructure, cautioned that much of the “hot money” rushing into the AI sector may not remain long enough to fully capture the long-term rewards of these investments.
“I’m taking it from my own infrastructure standpoint, but there is a lot of hot money in [AI] ... The problem that we have with that is that this theme is exciting. We love it, it’s long term, it’s great, but it’s not this year. It’s not next year, it’s three, four, five years before you get to see that data centre development, before you get to see that load development,” Shaw said.
“We don’t know how it’s going to translate to earnings. So is that hot money going to stick around expecting that it’s going to be realised?”
She highlighted the US is witnessing accelerated growth in data centre development, driving a surge in power load demand across various regions.
But, hyperscalers – large-scale data centre operators with power requirements exceeding 25 megawatts – are now facing wait times of two to seven years for incremental supply to meet their operational needs. This highlights significant challenges in scaling infrastructure to match rapid technological and digital expansion.
“The AI data centres that have gone gangbusters, they need a huge amount of energy. The forecasts are great, like 6, 8, 8 per cent,” Shaw explained. “But you’re not going to see 6–8 per cent this year. You’re not going to see it next year.”
In a recent outlook, sustainability portfolio manager Peter Aquilina highlighted the challenges of navigating future trends in the data centre sector, citing the rapid evolution of technological advancements and digital services as key complexities.
“While 4D is excited about the long-term infrastructure investment thematic underpinned by the anticipated growth in data centres, it’s worth noting that this growth in investment and or demand will not be immediately realised in numbers,” Aquilina wrote.
Expounding on this, he said it takes time to plan, locate and build data centres, and this time lag flows to the execution of generation capacity, grid connections and load demand.
“It also must be financed by investors through debt and/or equity,” Aquilina added.
“While supportive of valuations, shareholders must be patient in the realisation of the opportunity set.”
