Record revenue, halved profit, and a deliberate choice
New Zealand’s largest home-grown IT firm just posted the kind of result that reads badly on a wire and rather differently once you understand what it bought. Datacom Group’s revenue hit a record $1.584 billion for the year to 31 March 2026, up 6.8% from $1.483 billion, while net profit after tax fell 46% to $20 million from $37 million.
The rest of the numbers move the same way. EBITDA slipped to $133 million from $147 million, operating cash flow more than halved to $75 million from $164 million, and bank borrowings rose to $131 million to fund the expansion. On paper, alarming. In practice, this is what it costs to build ahead of demand.
Where the money went
The centrepiece is infrastructure. Datacom acquired T4’s East Auckland data centre at Highbrook during the year, taking its New Zealand data centre count to five, and is upgrading Highbrook for high-density computing and more energy-efficient operations. That is the specific kind of plant needed to run AI workloads rather than conventional enterprise IT, and it does not come cheap.
The company also pre-purchased components and warehoused stock to insulate customers from a chaotic supply environment. CEO Greg Davidson did not sugar-coat it: “This is the most volatile world I’ve seen for supply chain. It’s worse than Covid,” he told RNZ. Middle East conflict and surging global demand for AI hardware are compressing availability and pushing up prices at exactly the moment Datacom wants to buy.
Davidson framed the whole result as intentional. “Heading into the year, we knew conditions would be challenging,” he said, adding that the company made “deliberate investments that strengthen the foundations our customers depend on. Those decisions affect short-term profit, but they are the right choices for resilience and future growth.”
The sovereignty pitch
The strategic logic rests on data sovereignty. As Davidson told Reseller News, sovereign infrastructure gives organisations certainty over data governance and access, with Datacom building local foundations while staying connected to global platforms. For government agencies, regulated financial institutions and any business with data residency obligations, running AI through offshore hyperscale cloud carries compliance risk. Datacom wants to be the onshore answer.
And it is already shipping product, not slideware. Among the projects completed this year were AI-driven search for the Financial Markets Authority website and an AI simulator for a food production company – production deployments for regulated and industrial customers, not proof-of-concept demos.
The catch is the customers
Here is the tension. The buyers Datacom is selling this capacity to are cautious. Davidson acknowledged New Zealand customers remained wary after a downturn deeper than Australia’s, scrutinising every dollar of technology spend. “We’ve got very price-sensitive buying. We’ve got customers really looking for value in what’s being delivered,” he said. AI projects proceed only where there is a clear productivity payoff, not on speculation.
So Datacom is spending big to build capacity into a market that is still deciding whether it wants it. That is the gamble laid bare.
Why it can afford to wait
The reason a 46% profit drop is a strategy rather than a crisis comes down to ownership. Datacom is 55% owned by the Holdsworth family and 45% by the NZ Super Fund, both patient, long-horizon holders with no quarterly earnings treadmill to satisfy. A listed peer posting these numbers would be fielding hostile questions from analysts. Datacom gets to build through the gap.
With 5,935 employees across New Zealand, Australia and Southeast Asia, up from 5,375 a year earlier, it is also a major sector employer making a national-scale wager.
What it means for everyone else
The wider lesson for business decision-makers is not about one company’s margin. It is about the economics of AI infrastructure itself. The capital required to build AI-ready compute is enormous, the lag between spending and revenue is long, and only players with patient capital can absorb the gap. Any New Zealand business that wants to run serious AI workloads onshore faces that same wall – unless someone builds the capacity first. Datacom is effectively making that bet on the market’s behalf.
The open question is whether the domestic market is big enough, and whether demand arrives fast enough, to justify the spend before the borrowing bill compounds. Davidson has committed the capital. The next couple of years will tell us whether he read the demand curve right.