“In some cases, natural attrition has not been back-filled,” Davidson said.

Some professional services roles had been added to meet market demand for expertise in AI.

A “recessionary environment” resulted in a notable reduction in customer spending in New Zealand, Davidson said.

In the corporate and government sectors, “Many organisations had to push the ‘pause’ button on key technology projects”, while his firm’s payroll business revealed that many Kiwi small businesses had “a very difficult year”.

But continued gains across the Tasman, including “several notable contract wins, particularly in the government space” resulted in the Wellington-founded IT services, solutions and data centre operator lifting earnings overall.

As well as its more buoyant economy, surveys have found Australia ahead of NZ in AI adoption.

Datacom’s net profit for the year to March 31, 2025 was $37 million, up from FY2024’s $34m.

Revenue – which Davidson said was evenly balanced between Australia and New Zealand – increased from the year-ago $1.47 billion to $1.48b.

That makes Datacom – which is 55% owned by the rich-list Holdsworth family and 45% by the New Zealand Superannuation Fund – New Zealand’s fourth-largest tech firm by revenue for FY2025, according to the Technology Investment Network (TIN).

NZX-listed F&P Healthcare ($1.74b) is the largest, followed by ASX-listed Xero ($1.71b) and F&P Appliances ($1.51b; now owned by China’s Haier). The export-focused TIN doesn’t include telecommunications companies.

Operating cashflow rose to $164m from the year-ago $139m.

Net debt reduced from $82m to $45m.

‘Biggest revolution I’ve seen’

Datacom – which absorbed NZ Post’s IT division in 1991, with the NZ Super Fund taking over NZ Post’s stake in 2012 – turned 60 this year.

Davidson, who’s been with the firm for nearly half of that time, says he’s seen a welter of tech changes, but nothing like today.

“AI is the biggest revolution I’ve seen, maybe including the web. It’s certainly bigger than mobile,” he says.

“And every six months, what’s possible has just advanced at a rate like no other technology disruption I’ve seen.”

Davidson says the three stages of AI have been basic large language model (LLM) bots you can chat to, the emergence of “expert” AIs and now agents.

He says an example of an expert system is the AI assistant Datacom developed for its Datapay payroll solution, which you ask complex questions using natural language prompts.

After an in-house trial on its own staff, it will be released to the market in July.

“The large language models tend to be a little too friendly and helpful, and that includes making stuff up when they don’t know the answers,” Davidson says.

The Datapay AI assistant, by contrast, is an expert system because guardrails ”are sitting around it to ensure it’s very clear on what it can answer and what it can’t answer, and it stays only on the topic of payroll”.

Carefully curated data was drawn from “core agencies like IRD, MBIE and ACC, but case law was excluded because it can give non-deterministic [not certain] answers”, Davidson said.

“That’s a classic example of the kind of thinking that needs to go into building any expert based on a large language model. You’ve got to organise your data really carefully, and you’ve got to organise guardrails around how it operates in order to ensure that it achieves expert outcomes.

“The third stage of it is agents – and that’s basically large language models with code wrapped around them that can do things on your behalf and interact with each other,” Davidson says.

“That’s what we’ve been using for [software] coding automation and business analysis.”

Without naming clients, Datacom says its AI projects across the year have included:

With its neutral consultancy business, Datacom works with several big cloud, AI and hyperscale data centre players including Microsoft, AWS and Google.

But its horses-for-courses approach also has it operating four data centres of its own, and creating its own products.

“Over the past two financial years we have made significant investment around upgrading our data centres and building out our SaaS [software-as-a-service or cloud software] product capabilities, including the launch of our Water Intelligence Asset – or WAI – tool for local councils,” Davidson says.

Davidson expects conditions on this side of the Tasman to remain “economically challenging” over the next 12 months but says the artificial intelligence revolution will open new opportunities.

“AI is becoming fundamental to being efficient and competitive these days, so customers are needing a lot of hand-holding as to which tool to use for what job and where to do the processing to avoid the costs eroding the benefits.”

Chris Keall is an Auckland-based member of the Herald’s business team. He joined the Herald in 2018 and is the technology editor and a senior business writer.